1 INFO-VAX	Wed, 28 Jun 2000	Volume 2000 : Issue 358       Contents:N Re: (REPOSTED ) How to install SRM console from alpha firmware version 4.45 onM Re: (REPOSTED ) How to install SRM console from alpha firmware version4.45 on   Re: Charon-VAX Hobbyist Question  Re: Charon-VAX Hobbyist Question  Re: Charon-VAX Hobbyist Question  Re: Charon-VAX Hobbyist Question  Re: Charon-VAX Hobbyist Question Compaq advertizes  RE: Compaq advertizes & Re: Compaq paying for software ports ?# Current number of running processes ' Re: Current number of running processes ' Re: Current number of running processes ' Re: Current number of running processes ' RE: Current number of running processes ' Re: Current number of running processes " RE: DECNET-Plus, DTSS, NTP and UTC" Re: DECNET-Plus, DTSS, NTP and UTC" RE: DECNET-Plus, DTSS, NTP and UTC" Re: DECNET-Plus, DTSS, NTP and UTC* Re: Disponibility of OPenVMS on Vax System* Re: Disponibility of OPenVMS on Vax System* Re: Disponibility of OPenVMS on Vax System1 FA: Digital Pathworks For DOS V4.1 OpenVMS SEALED . Re: Found: PC package to make backup of VMS CD( Free + postage for Old VAX Hardware docs Re: General discussion comment Re: General discussion commentB Re: How to install SRM console from alpha firmware version 4.45...= RE: ideas for accessing common data between multiple clusters 7 Re: Northern Light vs. Google (and the winner is . . .) 7 RE: Northern Light vs. Google (and the winner is . . .) 7 Re: Northern Light vs. Google (and the winner is . . .) 7 Re: Northern Light vs. Google (and the winner is . . .) 7 Re: Northern Light vs. Google (and the winner is . . .) 7 Re: Northern Light vs. Google (and the winner is . . .) 7 Re: Northern Light vs. Google (and the winner is . . .) 7 Re: Northern Light vs. Google (and the winner is . . .)  ODS2 Reader/Writer( Re: OpenVMS Alpha C programming question( Re: OpenVMS Alpha C programming question OT - FEELING OLD!! Re: OVMS 7.2 updates Re: print to file 	 SMTP Mail 
 Re: SMTP Mail  SMTP Mail - followup! VAX based 1032 database to Oracle % Re: VAX based 1032 database to Oracle  Re: VAX on Intel? ; Re: what layered products can i install with these licences = Will OpenVMS/VAX 7.3 Include VMSINSTAL Installation History ?   F ----------------------------------------------------------------------   Date: 27 Jun 2000 19:46:02 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)W Subject: Re: (REPOSTED ) How to install SRM console from alpha firmware version 4.45 on 6 Message-ID: <8jb09q$bkm$1@mailint03.im.hou.compaq.com>  ^ In article <20000625230658.16399.00002189@ng-ft1.aol.com>, gigglegs@aol.com (Gigglegs) writes:C :  ...IS THERE ANY GOOD MANUAL THAT WILL HELP   SOME ONE LIKE ME TO O :INSTALL (OPEN VMS ON THIS MACHINE OR  GET TO SRM CONSOLE) .   ANY STEP BY STEP & :DETAILED HELP WOULD BE APPRECIATED...  I   As you are likely aware of what the uppercase convention means, please     don't yell.  Thanks!  H   Switching from AlphaBIOS or ARC to SRM is probably the weakest spot ofG   the existing AlphaStation documentation -- hence the inclusion of the D   information that I mentioned in an earlier into the OpenVMS FAQ...  H   It would appear that you are past the information on switching over toF   the SRM covered in the FAQ, and have moved along to another problem.  C   The stackdump you report is probably due to unsupported hardware  F   configuration.  Two common triggers: either insufficient (or failingG   or erroneous) memory (OpenVMS Alpha wants at least 32 MB) or graphics 3   controller that is not supported by the firmware.   H   When posting, please take the time to post the particular error(s) or G   message(s) you are seeing -- this information (usually) makes it far  D   easier to figure out what problem(s) you are encountering.  Again,1   thanks for your understanding in this matter...   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Tue, 27 Jun 2000 17:26:37 -0500 2 From: "Craig A. Berry" <craig.berry@metamorgs.com>V Subject: Re: (REPOSTED ) How to install SRM console from alpha firmware version4.45 onM Message-ID: <4.3.2.7.2.20000627160404.00c36100@exchi01.midwest.metamorgs.com>   
 Hoff said:  _ >In article <20000625230658.16399.00002189@ng-ft1.aol.com>, gigglegs@aol.com (Gigglegs) writes: A >: ...IS THERE ANY GOOD MANUAL THAT WILL HELP SOME ONE LIKE ME TO M >:INSTALL (OPEN VMS ON THIS MACHINE OR GET TO SRM CONSOLE) . ANY STEP BY STEP ' >:DETAILED HELP WOULD BE APPRECIATED...  > [ >As you are likely aware of what the uppercase convention means, please don't yell. Thanks!   L Yes, do avoid yelling when posting, but on the other hand I understand that M not being able to obtain a manual for a second-hand system is enough to make  G one want to scream.  Since no one has yet answered the manual question  M directly, I will mention that they will sell you the manual, apparently, for  * $45.  Look for part number EK-PCDTA-UI at:  2 <http://www.digital.com/CAS-Catalog/das07qhm.html>  I Systems with some similarities (but some differences) have their manuals  L published for free on the Web and there is no readily discernible rationale H for what's available.  The Alphaserver 400 manual describes a system of K about the same vintage and will have pretty much the same console commands:   C <http://www.compaq.com/alphaserver/archive/400/ek-pcsva-ui-a01.pdf>   K Note, however, that there are important differences that may or may not be  I misleading in tracking down your particular problem.  On the AS 400, for  H example, it is the J4 jumper on the motherboard that determines whether K console switching from ARC to SRM is disabled, whereas on the Alphastation  , 200 it is J6 that determines the same thing.  0 The SRM console is documented in more detail at:  k <http://www.support.compaq.com/alpha-tools/documentation/archive/sdk/ec-qk8df-te_srm_cons_mb_users_gde.pdf>   M In the lower right hand corner of the Alphastation 200 (when viewed from the  L rear) there is a row of 8 LEDs that indicate status.  The meanings of these F are the same as documented on page 4-8 of the AS 400 manual (see link K above).  You probably need memory, as folks have mentioned, but this might  5 confirm that or provide other clues if that's not it.     / _______________________________________________ 1 Craig A. Berry                                      mailto:craig.berry@metamorgs.com  , "Literary critics usually know what they're & talking about. Even if they're wrong." 		-- Perl creator Larry Wall   ------------------------------  # Date: Wed, 28 Jun 2000 02:34:14 GMT 2 From: "Zane H. Healy" <healyzh@shell1.aracnet.com>) Subject: Re: Charon-VAX Hobbyist Question 4 Message-ID: <Gud65.315$9e.67801@typhoon.aracnet.com>  . Hunter Goatley <goathunter@goatley.com> wrote:A > You can do it the way I'm doing it.  It's not as nice as having $ > Ethernet capability, but it works.  B > The VMS container files used by Charon-VAX are the same type of F > container files used by VDDRIVER or the LD virtual disk driver (see D > the OpenVMS freeware CD or ftp.wku.edu).  To move files in and outG > of the Charon-VAX environment, I simply created a VDDRIVER container  F > file called TRANSFER.DSK, copied the desired files to it, then FTPedH > the container file to my laptop, and loaded Charon-VAX with one of theF > MSCP drives specifying TRANSFER.DSK.  Voila.  A method, albeit a bitD > cumbersome, for getting your own files into and out of Charon-VAX.  I Most cool!  I'll have to give this a try!  While not as easy as I'd like, ! this should work for what I want.   L On the down side the tool for turning a CD into a disk image doesn't seem toI work on a Mac under Virtual PC, so I've not gotten it set up yet.  Once I L get it working under VPC I should be able to try it on an iBook, as I've gotK one to play with through next weekend :^)  Should be slightly faster than a  MicroVAX II.   			Zane    ------------------------------  # Date: Wed, 28 Jun 2000 02:35:43 GMT 2 From: "Zane H. Healy" <healyzh@shell1.aracnet.com>) Subject: Re: Charon-VAX Hobbyist Question 4 Message-ID: <3wd65.316$9e.67801@typhoon.aracnet.com>  H Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de> wrote:4 > Timothy Stark <sword7@grace.speakeasy.org> writes:  C >> No, it does not support Ethernet.  It is strictly closed system.   N > Hmmm, the "Specifications" web page says it emulates the DEQNA. Is it wrong?  I That's the full blown commercial version.  They disabled ethernet support J and some other stuff on the Hobbyist version.  Which unfortunatly makes it almost worthless for hobbyists.    			Zane    ------------------------------  + Date: Wed, 28 Jun 2000 02:53:42 +0000 (   ) 3 From: Christopher Smith <chriss@Mufasa.pubserv.com> ) Subject: Re: Charon-VAX Hobbyist Question J Message-ID: <Pine.LNX.4.05.10006280252260.10390-100000@Mufasa.pubserv.com>  ) On Wed, 28 Jun 2000, Zane H. Healy wrote:   N > On the down side the tool for turning a CD into a disk image doesn't seem toK > work on a Mac under Virtual PC, so I've not gotten it set up yet.  Once I N > get it working under VPC I should be able to try it on an iBook, as I've gotM > one to play with through next weekend :^)  Should be slightly faster than a  > MicroVAX II.  G Seems to me that a macintosh, emulating a windows box, emulating a vax, > might have so much overhead as to negate any speed increase...   Regards,   Chris   O =============================================================================== @ "My two cents"			(http://rootworks.com/twocentsworth.cgi?128562)= Christopher Smith(chriss@pubserv.com)			Prgramer^W Programmer  Prime Synergy of Champaign, IL. % ------------------------------------- I "Where a calculator on the ENIAC is equipped with 18,000 vacuum tubes and H weighs 30 tons, computers in the future may have only 1,000 vacuum tubes; and weigh only 1.5 tons." -- Popular Mechanics, March 1949  O -------------------------------------------------------------------------------    ------------------------------  # Date: Wed, 28 Jun 2000 04:01:32 GMT 2 From: "Zane H. Healy" <healyzh@shell1.aracnet.com>) Subject: Re: Charon-VAX Hobbyist Question 4 Message-ID: <wMe65.317$9e.68332@typhoon.aracnet.com>  4 Christopher Smith <chriss@mufasa.pubserv.com> wrote:I > Seems to me that a macintosh, emulating a windows box, emulating a vax, @ > might have so much overhead as to negate any speed increase...  L Theoretically a 300Mhz iBook should equal about 1.5 VUPS.  Note the key wordB is theoretically, I still need to get it working under Virtual PC.   			Zane    ------------------------------  # Date: Wed, 28 Jun 2000 04:06:21 GMT 0 From: Timothy Stark <sword7@grace.speakeasy.org>) Subject: Re: Charon-VAX Hobbyist Question 8 Message-ID: <1Re65.282179$VR.4160411@news5.giganews.com>  1 Zane H. Healy <healyzh@shell1.aracnet.com> wrote: K > That's the full blown commercial version.  They disabled ethernet support L > and some other stuff on the Hobbyist version.  Which unfortunatly makes it! > almost worthless for hobbyists.   F Yeah. Almost useless for Hobbyists.  I know that companies rather care$ about businesses than hobbyists. :-(  H I did e-mailed to SRI who developed Charon-VAX emulator last weekend.  IH was told that there will be *NO* hobbyist versions that support ethernetG in future.  Also future versions of Charon-VAX will requires Windows NT G and 2000.  I asked for a listing of prices.  I was told that it will be & prices above "home brand" prices.  :-(   -- Tim Stark   --  C Timothy Stark	<><	Inet: sword7@speakeasy.org, sword7@firesword7.net J --------------------------------------------------------------------------F "For God so loved the world, that he gave his only begotten Son, that H whosoever believeth in him should not perish, but have everlasting life.. Amen." -- John 3:16 (King James Version Bible)   ------------------------------    Date: 27 Jun 2000 15:11:47 -0500, From: koehler@eisner.decus.org (Bob Koehler) Subject: Compaq advertizes+ Message-ID: <M8FbNJS8i8lL@eisner.decus.org>   C   So who was it at Compaq that actually understood that they should A   make something of the Alphas used in the human geneome project?   (   Full page add today (Washington Post).  D   Does this spell the end of stealth marketing (is DEC really gone)?     Will they ever market VMS?   ------------------------------  % Date: Tue, 27 Jun 2000 20:23:55 -0400 + From: "Main, Kerry" <Kerry.Main@compaq.com>  Subject: RE: Compaq advertizesJ Message-ID: <910612C07BCAD1119AF40000F86AF0D80528449A@kaoexc4.kao.dec.com>  " Also, big ad ran in Boston Globe..   Note - Forbes article:3 http://www.forbes.com/tool/html/00/Jun/0626/mu7.htm   " <<< Will they ever market VMS? >>>  . Wait until tomorrows (June 28) announcements.   J Folks on this list should like what they hear .. well, ok maybe one or two folks might not ..   :-)    Regards,  
 Kerry Main Senior Consultant,
 Compaq Canada  Professional Services  Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.com        -----Original Message-----@ From: koehler@eisner.decus.org [mailto:koehler@eisner.decus.org]$ Sent: Tuesday, June 27, 2000 4:12 PM To: Info-VAX@Mvb.Saic.Com  Subject: Compaq advertizes      C   So who was it at Compaq that actually understood that they should A   make something of the Alphas used in the human geneome project?   (   Full page add today (Washington Post).  D   Does this spell the end of stealth marketing (is DEC really gone)?     Will they ever market VMS?   ------------------------------    Date: 27 Jun 2000 14:53:28 -0500, From: koehler@eisner.decus.org (Bob Koehler)/ Subject: Re: Compaq paying for software ports ? + Message-ID: <j6SmgJO+++Tj@eisner.decus.org>   R In article <8j8e8a$kq4$1@pyrite.mv.net>, "Bill Todd" <billtodd@foo.mv.com> writes: > J > I assume that even strings with descriptors get probed on a page-by-page1 > basis on entry to a normal VMS system service.    F  No.  Not required.  Any attempt to write will be limitted by the sizeE  given by the descriptor, and valid access to the page by the calling H  mode will be verified by memory management hardware if the CPU is doing
  the writing.   H  Page by page probing is done by the CPU prior to starting a DMA from an@  external device in a properly written device driver (built into+  standard routines almost all drivers use).    < How much more expensive M > (compared to the normal cost of entering, executing, and exiting a service) D > can it really be to scan for a null up to the next page boundary,   D  Byte access is expensive.  Probing for a single 0 byte burns up CPU  cycles.  E  But UNIX runs fine on an Alpha, which shows what you can do when you !  have all that CPU power to burn.    ------------------------------  % Date: Tue, 27 Jun 2000 20:16:21 +0200 ) From: "Tomer Cohen" <tomer@flashmail.com> , Subject: Current number of running processes0 Message-ID: <8janhp$3s8$1@news.netvision.net.il>   Hi, J Is there any easy way of knowing the number of processes currently runningG in the system other than putting all the show system information into a % temporary file and counting the rows?    Thanks,  -- Tomer Cohen  Tomer@FlashMail.com    ------------------------------  % Date: Tue, 27 Jun 2000 14:14:55 -0500 * From: "Mark E. Levy" <levy@sysman-inc.com>0 Subject: Re: Current number of running processes/ Message-ID: <slhv8jjvjev132@corp.supernews.com>   
 $ SHOW MEMORY    Loot at the "slot usage."      --E ---------------------------------------------------------------------  Mark E. Levy, President " System Management Associates, Inc.! 888-291-5055 x202 (Illinois Only) $ 847-291-1550 x202 (Outside Illinois) 847-291-3866 fax www.sysman-inc.com levy@sysman-inc.com E --------------------------------------------------------------------- 2 Tomer Cohen <tomer@flashmail.com> wrote in message* news:8janhp$3s8$1@news.netvision.net.il... > Hi, L > Is there any easy way of knowing the number of processes currently runningI > in the system other than putting all the show system information into a ' > temporary file and counting the rows?  > 	 > Thanks,  > --
 > Tomer Cohen  > Tomer@FlashMail.com  >  >    ------------------------------  # Date: Tue, 27 Jun 2000 19:35:14 GMT ( From: Alan <campbela_691021@my-deja.com>0 Subject: Re: Current number of running processes) Message-ID: <8javlh$qjv$1@nnrp1.deja.com>M  0 In article <8janhp$3s8$1@news.netvision.net.il>,,   "Tomer Cohen" <tomer@flashmail.com> wrote: > Hi,uD > Is there any easy way of knowing the number of processes currently running G > in the system other than putting all the show system information into  a ' > temporary file and counting the rows?n >u	 > Thanks,n > --
 > Tomer Cohenu > Tomer@FlashMail.com  >c >e   Try :-  . PIPE SHOW SYS | SEARCH SYS$INPUT ""/NOOUT/STAT  C That should give you a count of the number of lines in the show syse5 output (you may have to discount the header details).,   Alan -- ---C
 Alan Campbellu alanc@mullen.demon.co.uk    & Sent via Deja.com http://www.deja.com/ Before you buy.d   ------------------------------  + Date: Tue, 27 Jun 2000 17:02:00 -0300 (EST)  From: becherini@vortex.ufrgs.brO0 Subject: Re: Current number of running processes, Message-ID: <00062717015952@vortex.ufrgs.br>  : Received:	by vortex.ufrgs.br (V5.0A-1, OpenVMS V7.2 Alpha)+ From:		Fabio Becherini <becherini@ufrgs.br>e Reply-to:	<becherini@ufrgs.br>< Comments:	@vortex.ufrgs.br, vortex(46.451)::, psi%........::2 References:	BR, TCHE, UFRGS, CPD network, Cia-INFO- Organization:	Cia-INFO /DRS /CPD-UFRGS /UFRGSeO _______________________________________________________________________________   , . From: "Mark E. Levy" <levy@sysman-inc.com>2 . Subject: Re: Current number of running processes' . Date: Tue, 27 Jun 2000 14:14:55 -0500  . To: Info-VAX@Mvb.Saic.Come .  . $ SHOW MEMORYG .  . Loot at the "slot usage."  .  .  . --G . ---------------------------------------------------------------------  . Mark E. Levy, President $ . System Management Associates, Inc.# . 888-291-5055 x202 (Illinois Only)l& . 847-291-1550 x202 (Outside Illinois) . 847-291-3866 fax . www.sysman-inc.com . levy@sysman-inc.com G . --------------------------------------------------------------------- 4 . Tomer Cohen <tomer@flashmail.com> wrote in message, . news:8janhp$3s8$1@news.netvision.net.il... . > Hi,cN . > Is there any easy way of knowing the number of processes currently runningK . > in the system other than putting all the show system information into aJ) . > temporary file and counting the rows?z . >c . > Thanks,  . > -- . > Tomer CohenE . > Tomer@FlashMail.comR     	how about MONITOR SYSTEM ?s  	 	regards,   N  _____________________________________________________________________________O |                                                                             |TO | Fabio Becherini                   System & Network Manager, Webmaster UFRGS | O | CPD-UFRGS                         Centro de Processamento de Dados da UFRGS |BO |                                   Universidade Federal do Rio Grande do Sul |aO |                                                   Divisao de Rede e Suporte |iO |                                          (55)(51) 316-5041 / 331-1215 (fax) |eO | Rua Ramiro Barcelos, 2574  -  Santa Cecilia  -  Porto Alegre - RS -  Brasil |nO |_____________________________________________________________________________|eO |                                                                             | O | Cia-INFO (c) Ophicin@ das Informacoes                   Coordenacao Central |oO |_____________________________________________________________________________|(O |                                                                             |tO | INTERnet:  fabio.becherini@ufrgs.br              DECnet:  vortex::becherini |aO |_____________________________________________________________________________|t   ------------------------------  % Date: Tue, 27 Jun 2000 17:29:36 -0400,+ From: "Main, Kerry" <Kerry.Main@compaq.com>t0 Subject: RE: Current number of running processesJ Message-ID: <910612C07BCAD1119AF40000F86AF0D805284496@kaoexc4.kao.dec.com>   Tomer ..   Try one of these approaches:  ) $ PIPE MON SYS | SEARCH SYS$INPUT "total"29 [subtract 2 for sub-processes spawned for these commands]e   or a  I $ monitor system (gives total processes plus breakdown of process states)r   Regards,  
 Kerry Main Senior Consultant,
 Compaq Canada0 Professional Servicesm Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.comn       -----Original Message-----. From: Tomer Cohen [mailto:tomer@flashmail.com]$ Sent: Tuesday, June 27, 2000 2:16 PM To: Info-VAX@Mvb.Saic.ComI, Subject: Current number of running processes     Hi,:J Is there any easy way of knowing the number of processes currently runningG in the system other than putting all the show system information into av% temporary file and counting the rows?h   Thanks,n -- Tomer Coheno Tomer@FlashMail.comn   ------------------------------  % Date: Tue, 27 Jun 2000 16:58:42 -0500 ) From: "John E. Malmberg" <wb8tyw@qsl.net>u0 Subject: Re: Current number of running processes/ Message-ID: <sli8gj4ljev153@corp.supernews.com>k  + Tomer Cohen <tomer@flashmail.company> wrote.5 in message news:8janhp$3s8$1@news.netvision.net.il...s  L > Is there any easy way of knowing the number of processes currently runningI > in the system other than putting all the show system information into al' > temporary file and counting the rows?e   For VMS 6.2 and later:   $SHOW SYSTEM/NOPROC/TOTALe  6 Or you can use f$pid() to loop through and count them.   -Johna wb8tyw@qsl.network   ------------------------------   Date: 27 Jun 2000 13:28:47 PDTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)+ Subject: RE: DECNET-Plus, DTSS, NTP and UTC 3 Message-ID: <QGIwUvHdhTSO@mccdev.slac.stanford.edu>o  E In article <C15945A9D9EFCF11BA8B08002BBF1CCC0CD735@berry.mvpsi.com>,  *     	John Vottero <John@MVPSI.com> writes:K > I thought that I had finally figured this out this past April.  It sounds.N > like I'm going to be pulling out more hair this October!  Are you using UCX?           Nope, Multinet 4.2A.  . > Do you have the time zone configured in UCX?  H         No and yes.  Not UCX, but it is configured in Multinet.  I don'tD     think there is any interactino between Multinet and DTSS though.  M >                                               The time zone stuff in UCX issM > deprecated but that may be where I get my time zone after I delete the DTSS_ > file.  (That's just a SWAG).           Thanks, Ken  --  M  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edut:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------   Date: 27 Jun 2000 13:30:34 PDTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)+ Subject: Re: DECNET-Plus, DTSS, NTP and UTCb3 Message-ID: <tkyTQOaqC2mn@mccdev.slac.stanford.edu>r  . In article <39580645.47DDB85C@earthlink.net>, >     	"David J. Dachtera" <djesys.nospam@earthlink.net> writes:H >> In article <C15945A9D9EFCF11BA8B08002BBF1CCC0CD734@berry.mvpsi.com>, *     	John Vottero <John@MVPSI.com> writes: >> > ...then execute >> > UTC$CONFIGURE_TDF.COM.e >> r > ; > "Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515" wrote:d- >>     then  I  executed  UTC$TIME_SETUP.COM.e > ) > Note the differences. Just a thought...l  H         I'm trying to be ever-so-correct.  The "Hoff" always says "don'tH     run UTC$CONFIGURE_TDF.COM directly, use UTC$TIME_SETUP.COM instead,"H     so I did.  On my cluster running  DECNET  IV (no DTSS), I do in factH     use  UTC$CONFIGURE_TDF,  but that's not germain to  this  particular     discussion.            Thanks, Kent -- cM  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edug:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------   Date: 27 Jun 2000 13:41:36 PDTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)+ Subject: RE: DECNET-Plus, DTSS, NTP and UTCN3 Message-ID: <RKoI+2eN9n9N@mccdev.slac.stanford.edu>e  H In article <Pine.OSF.4.21.0006270703450.15548-100000@athena.csdco.com>, /     John Nebel <nebel@athena.csdco.com> writes:i  I > One has to start and stop DTSS after the semi-annual time change to setbG > the contents of a file used by a task run as a part of system startupsH > (DTSS$SET_TIMEZONE).  See SYS$COMMON:[SYS$STARTUP]DTSS$UTC_STARTUP.COM  H         I think this is the  key.   I  _thought_  I had disabled all theH     DTSS  stuff  by putting an EXIT at the top of  DTSS$STARTUP.COM  (inH     SYS$STARTUP:) which  otherwise  executes  DTSS$UTC_STARTUP.COM.   InH     fact,   DTSS$UTC_STARTUP.COM   is   also   executed   directly  fromH     NET$STARTUP.COM during system startup (under which circumstances I'm'     not sure, I didn't trace the flow).w  H         As an experiment, I put  an  EXIT at the top of DTSS$UTC_STARTUPH     as well and rebooted.  Sure enough, _only_ SYS$TIMEZONE_DIFFERENTIALH     gets defined (like on my DECNET IV nodes), however DECNET (Plus) didH     not  start  correctly  when  I did that.  But it  does  verify  thatH     SYS$SYSTEM:DTSS$SET_TIMEZONE.EXE, run from DTSS$UTC_STARTUP, is  theH     program  which  defines  the various SYS$TIMEZONE* logical names and>     is, apparently, the locus where PST vs. PDT is determined.  H         For now, I'm resigned to  letting  the DTSS Clerk start (all theH     DTSS  defaults)  having  blocked the "Too Few Servers"  messages  inH     NET$EVENT_LOCAL.NCL.  This works for me because there are _no_  DTSSH     Servers  on any of my networks and so the Clerk lets NTP control theH     time.  It will be interesting to see  what happens on the morning of     October 29...-  )         Thanks for all the responses, Kena -- eM  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edu>:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  # Date: Wed, 28 Jun 2000 02:53:52 GMT + From: Jonas Lindholm <jlindholm@nyc.rr.com>t+ Subject: Re: DECNET-Plus, DTSS, NTP and UTCV* Message-ID: <395968B7.3E7145B9@nyc.rr.com>  P I always let DTSS clerk be created and enabled to let it setup correct time zoneR information. I then disable the DTSS and use NTP to syncronize the time. This work
 ok for me.  . I add the line DISABLE DTSS at the end of file& SYS$STARTUP:NET$DTSS_CLERK_STARTUP.NCL  B To change between standard and daylight saving time I use the file% UTC$CONFIGURE_TDF.COM in a batch job.r   /Jonas Lindholmu      9 "Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515" wrote:t  / > In article <39580645.47DDB85C@earthlink.net>,eC >         "David J. Dachtera" <djesys.nospam@earthlink.net> writes: I > >> In article <C15945A9D9EFCF11BA8B08002BBF1CCC0CD734@berry.mvpsi.com>,l/ >         John Vottero <John@MVPSI.com> writes:  > >> > ...then execute > >> > UTC$CONFIGURE_TDF.COM.u > >> > >s= > > "Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515" wrote:h/ > >>     then  I  executed  UTC$TIME_SETUP.COM.  > > + > > Note the differences. Just a thought...t >wJ >         I'm trying to be ever-so-correct.  The "Hoff" always says "don'tJ >     run UTC$CONFIGURE_TDF.COM directly, use UTC$TIME_SETUP.COM instead,"J >     so I did.  On my cluster running  DECNET  IV (no DTSS), I do in factJ >     use  UTC$CONFIGURE_TDF,  but that's not germain to  this  particular >     discussion.- >- >         Thanks, Kena > --O >  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.EduC< >  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924< >  Menlo Park, CA  94025           |  FAX:      650-926-3515P >  -----------------------------------------------------------------------------D >  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------   Date: 27 Jun 2000 20:09:42 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)3 Subject: Re: Disponibility of OPenVMS on Vax SystemK6 Message-ID: <8jb1m6$bkm$2@mailint03.im.hou.compaq.com>  i In article <HXM55.21287$NS6.301741@news.globetrotter.net>, "jerome souc" <souje@globetrotter.net> writes:eH :I would like to see when Compaq stop the support and the availablity of :OPenVMS on VAX Systems.  F   I will assume that this was a request for information on when CompaqH   might stop support for OpenVMS VAX, and not (as it could also be read)G   a desire to see such support for OpenVMS VAX end.  (English is such a,?   fun language. :-)   If this asumption is correct, please see:A  *     http://www.compaq.com/alphaserver/vax/  /   And particularly, please the last section of:e  6     http://www.compaq.com/alphaserver/vax/vax_now.html  E   In particular: "Compaq Services intends to continue supporting VAX mD   customers through the year 2010. As stated in the customer letter,J   Compaq will continue to provide a full range of preventive, performance @   optimization, and remedial services for VAX/OpenVMS systems.".  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Tue, 27 Jun 2000 15:39:18 -0700i* From: "Jack Peacock" <peacock@simconv.com>3 Subject: Re: Disponibility of OPenVMS on Vax System1; Message-ID: <B2a65.436$h8.45582@news-west.usenetserver.com>   ? "Hoff Hoffman" <hoffman@xdelta.zko.dec.nospam> wrote in messagee0 news:8jb1m6$bkm$2@mailint03.im.hou.compaq.com... >vF >   In particular: "Compaq Services intends to continue supporting VAX$ >   customers through the year 2010. >aB Does anyone recall when DEC finally stopped support on TOPS-10 andG TOPS-20?  And how long that was after the DEC 10 and 20 were officiallyp@ discontinued?  How did it compare to their announced intentions?    Jack Peacock H (whose first systems programming assignment  was to make TRAFFIC-20 work with HP terminals)   ------------------------------  # Date: Wed, 28 Jun 2000 01:59:19 GMTx$ From: Ed Wilts <ewilts@mediaone.net>3 Subject: Re: Disponibility of OPenVMS on Vax System , Message-ID: <39595BF7.E9192C0D@mediaone.net>   Hoff Hoffman wrote:  > k > In article <HXM55.21287$NS6.301741@news.globetrotter.net>, "jerome souc" <souje@globetrotter.net> writes:CJ > :I would like to see when Compaq stop the support and the availablity of > :OPenVMS on VAX Systems. > H >   I will assume that this was a request for information on when CompaqJ >   might stop support for OpenVMS VAX, and not (as it could also be read)I >   a desire to see such support for OpenVMS VAX end.  (English is such a A >   fun language. :-)   If this asumption is correct, please see:r > , >     http://www.compaq.com/alphaserver/vax/  @ You do have to wonder about that URL though...  it's nice hiding5 VAX-specific information in an alphaserver directory..  P >  --------------------------- pure personal opinion ---------------------------N >    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com  $ 	.../Ed (proud to be running Vaxes!) -- - Ed Wilts Mounds View, MN, USA mailto:ewilts@mediaone.net   ------------------------------  # Date: Tue, 27 Jun 2000 19:16:23 GMTn5 From: "headley sappleton" <hsappleton@sprintmail.com>9: Subject: FA: Digital Pathworks For DOS V4.1 OpenVMS SEALEDD Message-ID: <b4765.4541$_b3.182400@newsread1.prod.itd.earthlink.net>  @ http://cgi.ebay.com/aw-cgi/eBayISAPI.dll?ViewItem&item=367988596G Auction Ends on:                 Thursday, Jul 06, 2000 at 09:12:15 PDTm   ------------------------------  + Date: Tue, 27 Jun 2000 19:23:26 -0500 (CDT)o From: sms@antinode.org7 Subject: Re: Found: PC package to make backup of VMS CDn) Message-ID: <00062719232614@antinode.org>i  C From: Eberhard Heuser-Hofmann <vaxinf@chclu.chemie.uni-konstanz.de>03 > The same (and free) solution exists for vms, see .- > ftp://v36.chemie.uni-konstanz.de/diskreadero  F    I normally use MOUNT /FOREIGN and COPY to dump a CD-ROM to a file. D There's an annoying error message at the end, but it always seems toB work.  The CDRECORD package includes READCD to do the same thing. B (Before I found "readcd" for the SunOS 5 system at work, we used aA VAXsta 3100 with the DCL method for all our CD-ROM copying needs,hH including bootable Sun Solaris distributions.  I gather that "dd" can be( made to do it on a Unix system as well.)  F    Does anyone know (that's _know_) of any problems using any of these methods?  H ------------------------------------------------------------------------  C    Steven M. Schweda               (+1) 651-699-9818  (voice, home)oC    382 South Warwick Street        (+1) 763-781-0308  (voice, work)-G    Saint Paul  MN  55105-2547      (+1) 763-781-0309  (facsimile, work)t9    sms@antinode.org                sms@provis.com  (work)s   ------------------------------  % Date: Tue, 27 Jun 2000 13:25:11 -0700g+ From: "Barry Treahy, Jr." <Treahy@mmaz.com>b1 Subject: Free + postage for Old VAX Hardware docse( Message-ID: <39590DA7.CDF6766E@mmaz.com>  H I'm purging my shelves of nostalgic materials.  Unless someone speaks upH quickly that they are interested, this listed items will be entering the circular file today.  6 VAX-11 Architecture Reference Manual - EK-VAXAR-RM-0010 VAX 11/780 Installation Manual - EK-SI780-IN-002@ VAX 11/785 Hardware Users Guide - EK-11785-UG-001 (two of these); VAX 11/780 Diagnostic System User's Guide - EK-DS780-UG-002R. DHU11 Interface User's Guide - EK-DHU11-UG-001" DZ11 Users Guide - EK-DZ110-UG-002$ DMF32 User's Guide - EK-DMF32-UG-002$ CI780 User's Guide - EK-CI780-UG-0010 SC008 Star Couple User's Guide - EK-SC008-UG-001+ HSC50 Installation Manual - EK-HSC50-IN-001i" HSC50 User Guide - EK-HSC50-UG-001  BA11-K Mounting Box Users Manual PDP-11 Bus Handbook-  G This is about 15 pounds for paper (all or nothing) and using USPS, that  is about $17.50 from AZ to MI.  5 The first person to say that they want it, gets it...    Barry    ------------------------------  % Date: Tue, 27 Jun 2000 14:20:55 -0400 ' From: "Bill Todd" <billtodd@foo.mv.com>t' Subject: Re: General discussion commente( Message-ID: <8jar8r$n5g$1@pyrite.mv.net>  > Fred Kleinsorge <kleinsorge@star.zko.dec.com> wrote in message% news:8j889t$m3p$1@lead.zk3.dec.com...rD > Be assured that there is a lot of thought/discussion on the topic. RMS/ODS_H > was a great framework for standardizing file formats for RSX, and then VMS.G > But the UNIX and NT "cached" file systems make VMS look poor - mostlyrI > because many times the file IO on these systems never hit the disk evera (forL > instance, the UNIX habits of creating and deleting temporary files).  Were- > UNIX forced to hit the disk for each write,   G Which is wholly unnecessary (as RMS buffering of sequential file writes I proves), unless the application actually depends on it (in which case, asiK with the use of $FLUSH with RMS, it's reasonable for the application to say_ so explicitly)._  !  and protect the integrity of the_+ > file system in the face of a system crash   F Which decent Unix file systems already ensure, by group-logging and/orH ordering writes rather than forcing all meta-data updates synchronously.  !  - then we wouldn't look too bad._B > Unfortunately you don't easily get the option of asking for less/ > reliability, and faster performance on VMS IO   H As I said, 'less reliability' is not at issue.  The unfortunate thing isB that you don't easily get the option of asking for faster file I/OH performance on VMS, period - and *some* performance features just aren't available at all.     - short of creating a RAMDISK.pJ > Even cached IO will still end up doing slow things when it has to change the7 > on-disk structure,  A Unless you create a file system that can handle such changes bothc7 efficiently and safely, as some Unix environments have.r   - bill   >r >s > & > Jan Vorbrueggen wrote in message ...L > >It seems to me that the crucial point is "Tru64 with its more lightweightC > >filesystem": that implies, to me, that using RMS naively in thatn
 > applicationoK > >is a problem, especially if you're programming via, say, the C RTL. In a 3 > >sense, you're buying more than you've asked for.t > >e > > Jani >o >    ------------------------------   Date: 27 Jun 2000 20:36:35 GMT2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)' Subject: Re: General discussion commentm, Message-ID: <8jb38j$r2n@gap.cco.caltech.edu>  R In article <8jar8r$n5g$1@pyrite.mv.net>, "Bill Todd" <billtodd@foo.mv.com> writes:I >As I said, 'less reliability' is not at issue.  The unfortunate thing is-C >that you don't easily get the option of asking for faster file I/O I >performance on VMS, period - and *some* performance features just aren'tr >available at all. >g  > - short of creating a RAMDISK.  J And even if you do that you only obtain performance parity for binary fileH IO.  If you need to move text files (records) it will be 2.5 - 6.5 timesC slower on OpenVMS than on Unix.  That's from some actual tests, seepB the thread:  ramdisk vs. file cache, and the winner is, file cache  5   http://www.deja.com/getdoc.xp?AN=633650884&fmt=textY  * Clearly there is a LOT of overhead in RMS.  K >> Even cached IO will still end up doing slow things when it has to change- >the >> on-disk structure,n >0B >Unless you create a file system that can handle such changes both8 >efficiently and safely, as some Unix environments have. >T  I OpenVMS can place binary blocks onto the disk as fast as, or faster than aK Linux.  It's all of the additional action involved in record handling that  K bogs down RMS, and the lack of automatic file caching that really eats into I performance.  We know that OpenVMS engineering is working on the caching aH problem, but the only stuff I saw in the roadmap about RMS concerns fileH locking, and it just doesn't seem reasonable that that would account forJ a 3X decrease in performance relative to Linux, in memory to memory sorts  of operations.  J I've asked this before and had no good answer - what the heck is RMS doing that eats so much time?    Regards,   David Mathog mathog@seqaxp.bio.caltech.eduf? Manager, sequence analysis facility, biology division, Caltech     ------------------------------   Date: 27 Jun 2000 19:03:16 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)K Subject: Re: How to install SRM console from alpha firmware version 4.45...A6 Message-ID: <8jatpk$ar6$1@mailint03.im.hou.compaq.com>  ^ In article <20000625001437.18040.00001570@ng-fn1.aol.com>, gigglegs@aol.com (Gigglegs) writes:L :ALpha station 200 4/233 system.  I tried putting in the hobbyist cd rom andK :tried installing new firmware on my system.  It doesn't recognize the fileAL :system.  I am new to this and I don't know what to do...any ideas?  My mainO :goal is to actually install open vms to my machine ( which is running winnt4.0>4 :right now).  Please any help will be appreciated.    E   If you are running Windows NT, then you have the ARC (or AlphaBIOS).(   console firmware loaded and running...  G   If so, you must go through the console set-up screens to switch over  F   to the SRM firmware -- please see the OpenVMS FAQ for some pointers F   to relevent information.  The FAQ section that is entitled "MGMT29. F   How do I switch between AlphaBIOS/ARC and SRM consoles?" will be of    particular interest here.T  =   The OpenVMS FAQ is available via a link located at the URL:   "     http://www.openvms.compaq.com/    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Tue, 27 Jun 2000 14:08:21 -0400S+ From: "Main, Kerry" <Kerry.Main@compaq.com>5F Subject: RE: ideas for accessing common data between multiple clustersJ Message-ID: <910612C07BCAD1119AF40000F86AF0D805284494@kaoexc4.kao.dec.com>   Alan,e  J Given that Customers with OpenVMS clusters and applications based on TCPIPK DNS (or DECnet) can do scheduled system maintenance (hw/os upgrades, tuningfJ reboots, patch upgrades etc) with ZERO application availability impact (noD failover required), I guess my first thought would be "why are these0 clusters being brought down in the first place?"  = Are there application specific issues here that require this?e   Regards,  
 Kerry Main Senior Consultant,
 Compaq Canadat Professional Services  Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.coms       -----Original Message-----/ From: Alan [mailto:campbela_691021@my-deja.com]C% Sent: Tuesday, June 27, 2000 10:56 AM  To: Info-VAX@Mvb.Saic.ComlB Subject: ideas for accessing common data between multiple clusters    D We have 2 Alpha clusters (may be increased to 3 in the near future).E Cluster 1 consists of 4 nodes which run a number of warehouse systemseC (in house).  These warehouse systems share some data by accessing awG directory that has been made available to them all.  Cluster 2 consists E of 2 nodes (one is used for H.A.) and a single warehouse system which F also needs access to the global area shared by all the other warehouseF system.  We have achieved this accessing the shared area and its files  across DECNET, which works fine.  @ The reason that we have a single warehouse on its own cluster is= because this warehouse needs 'up' a lot longer than the otherhC warehouses on cluster 1.  However taking down cluster 1 still meanseE that we can not get access to this shared data.  Does anyone have anymH ideas how we can limit the downtime on cluster 2 when cluster 1 is down.  D One thing that occured to me was somehow using shadowing (dont quiteF know the ins and outs of it) to create a temporary working copy of theG data while cluster 1 was unavailable, and getting it to 'catch-up' wheniG the cluster was available.  We currently use disk shadowing for backup..E We shutdown the warehouse systems, split the shadow sets and start-upr< the warehouse systems again (takes about 2 minutes).so takin   Any ideas ?a   Thanks   Alan ---o
 Alan Campbell- alanc@mullen.demon.co.uk    & Sent via Deja.com http://www.deja.com/ Before you buy.-   ------------------------------  % Date: Tue, 27 Jun 2000 14:11:15 -0400r' From: "Bill Todd" <billtodd@foo.mv.com>O@ Subject: Re: Northern Light vs. Google (and the winner is . . .)( Message-ID: <8jaqmp$msv$1@pyrite.mv.net>  4 Main, Kerry <Kerry.Main@compaq.com> wrote in messageD news:910612C07BCAD1119AF40000F86AF0D80528448B@kaoexc4.kao.dec.com... >  > mmm.. really mind boggling   That it is.c   >tD > >>> With 10,000 servers expected to be installed by the end of the	 year..<<<m   Now begins the FUD..   > * > - 10,000 servers - how are backups done?  G Primarily via the replication, I'd guess:  when you don't have to worrydL about user screw-ups (almost all of this is read-only content with automatedJ update), the need for backups is reduced to covering software screw-ups if your data is safely replicated.4  F If you're really worried, you capture the incoming update stream as itH arrives, in a manner that allows you to recover anything lost:  still noG backups to do, just restore if you need it.  But since at worst you can F always reconstruct the data anyway from the Web (which in any event is0 happening all the time), you may just not worry.  ; > - 10,000 servers - how are OS upgrades and patches done ?   ? Since the servers are mirrored, presumably by rolling upgrades.r  L > - 10,000 servers - interesting security, data integrity (virus checking ?) > challenges  F Again, no external party does any updating, so I'm not sure what those6 challenges might be (save for having a good firewall).  G > - 10,000 servers - but does not want expensive proprietary servers? IsJ > suspect NorthernLight and AltaVista may use less than 50-100 each. Would beL > interesting math question when one considers expensive computer room space > A/C etc ..  = Legitimate question, but those 1U boxes stack pretty tightly.-  3 > - 10,000 servers - likely all in one datacenter ?-   Not:  read the article.-    What if datacenter gets) > taken out by fire? What is DR solution?'I > - 10,000 servers - interesting financial challenges ie. depreciation byrJ > finance depts. is usually done over three year period, but this presents  > some interesting challenges ..  0 Not unless you depreciate them by hand on paper.  G > - 10,000 servers - what is their upgrade plan to new systems 12-18-24o months > from now?   L Again, rolling upgrade should work just as well as it does for VMS - if/when they decide they need it..   - bill   >s > To each his own .. >s > :-)n >m
 > Regards, >o >q > Kerry Main > Senior Consultant, > Compaq Canadat > Professional Services  > Voice : 613-592-4660 > FAX   : 819-772-7036 > Email : kerry.main@compaq.coma >  >  >f > -----Original Message-----. > From: Bill Todd [mailto:billtodd@foo.mv.com]& > Sent: Monday, June 26, 2000 11:43 AM > To: Info-VAX@Mvb.Saic.ComtB > Subject: Re: Northern Light vs. Google (and the winner is . . .) >o >q >r7 > Rob Young <young_r@eisner.decus.org> wrote in messageh' > news:VZkzGzloNz$H@eisner.decus.org...t >r > ..., > " > > Follow and read this overview: > >n > >  >vL http://www.intel.com/ebusiness/estrategies/casestudies/snapshots/google_p.ht > m. > > F > > They have many clusters, each apparently duplicated.  Each cluster@ > > contains partitioned data.  The 1000 chickens versus 64 oxen2 > > are winning.  But more about this in a bit :-) > H > Not bad.  Kind of surprises me that each uniprocessor Linux system can onlyL > handle 22 - 40 GB of data (with 256 MB - 1 GB of RAM), but it sounds as ifJ > they're limited by the choice of 100 Mbit Ethernet as the interconnect - andcL > the cost of anything faster (plus the multi-PCI, perhaps SMP system neededG > to make good use of it) may be enough higher than a vanilla commodityX system > to discourage that route.. >aJ > With the newer IDE disks supporting disconnectable operation (and GoogleK > already willing to write software, so they could write the IDE drivers touF > support this), it becomes even more attractive from the storage costB > viewpoint - and it sounds as if the cost of the IDE drives alone constitutesHD > close to 50% of the cost of each box.  Given this partitionable anI > application, it's almost impossible to compete on purchase price with aiI > larger centralized box using SCSI drives (leaving aside the question ofe why ? > IDE drives are so much less expensive); given that it's not aw
 life-criticalnI > application, the fact that the hardware is less self-checking and mighteJ > return brief garbage on the occasional failure isn't that important (andF > Linux is pretty stable on the OS end); given the box standardizationL > throughout the structure, sounds as if any failure results in an immediate4 > box-swap, and can then be trouble-shot at leisure. >XI > So the only obvious large-box argument would be the spaghetti of cablesrJ > linking the boxes.  Telcos lived with that problem for a very long time, so  > Google may be able to as well. >. > - bill >i > >  > > Robf > >f >a >n   ------------------------------  % Date: Tue, 27 Jun 2000 17:58:24 -0400n+ From: "Main, Kerry" <Kerry.Main@compaq.com>H@ Subject: RE: Northern Light vs. Google (and the winner is . . .)J Message-ID: <910612C07BCAD1119AF40000F86AF0D805284497@kaoexc4.kao.dec.com>   Bill,    >>> Now begins the FUD.<<<  & Sigh .. I'll ignore your pleasantries.  J >>> If you're really worried, you capture the incoming update stream as itB arrives, in a manner that allows you to recover anything lost: <<<  / For 10,000 servers with individual disk drives?t  B >>>> - 10,000 servers - how are OS upgrades and patches done ?<<<<G >>> Since the servers are mirrored, presumably by rolling upgrades. <<<t  E Now there's a good career for a few people .. start at server#1, when0F finished on server #10,000 start on server#1. Of course, this does notL consider FW updates, or any of the other issues associated with revision andH configuration management ...like driver versions, patches ect especially( since different server vendors are used.  K >>> > - 10,000 servers - but does not want expensive proprietary servers? IOK suspect NorthernLight and AltaVista may use less than 50-100 each. Would be J interesting math question when one considers expensive computer room space A/C etc ..<o  A >>Legitimate question, but those 1U boxes stack pretty tightly.<<   I Ok, even if the servers were $1k each (2 drives + 128Mb memory(?) + 100MbgI NIC+ additional racking charges), we are still talking about $10M for the K servers alone and this does not count the network switches and routers. Out'J of EACH cabinet, there would be 80 cables for NC connections. Of course, aJ good portion of the network would have to be redundant as well, so the keyH routers would be duplicated. 10,000 servers means 10,000 100Mb ports, soI this would mean approx 150x 64 port or 75x128 port switches - Can someone4I comment on how much these cost ? Are these costs factored into the 10,000 
 server costs?.  7 >>>>> - 10,000 servers - likely all in one datacenter ?,   >>> Not:  read the article.<<<  G Oops. Here you are right. I missed the text on different facilities. ItAK would be interesting to find out if these are duplicate facilities ie. eachs3 can function on its own, but thats for another day.   H >>> Again, rolling upgrade should work just as well as it does for VMS -# if/when they decide they need it.<<i  L Except that with OpenVMS cluster common system disks, upgrades are only doneJ once and when the other roots are rebooted, every system is now at the newF version. Big difference when you have to plan 10,000 individual system0 upgrades - even if using a streamlined approach.  L It appears to work for them. However, I suspect they are looking at HW costsL only and not the Operations and non-system related costs of maintaining suchH a complex environment. When datacenter space, configuration and revisionI mgmt, UPS, air conditioning, network, upgrade planning and implementation F costs are considered, I suspect the story might be a little different.   Again - to each his own. s   :-)n   Regards,  
 Kerry Main Senior Consultant,
 Compaq Canada- Professional Services  Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.comc       -----Original Message-----, From: Bill Todd [mailto:billtodd@foo.mv.com]$ Sent: Tuesday, June 27, 2000 2:11 PM To: Info-VAX@Mvb.Saic.Comn@ Subject: Re: Northern Light vs. Google (and the winner is . . .)      4 Main, Kerry <Kerry.Main@compaq.com> wrote in messageD news:910612C07BCAD1119AF40000F86AF0D80528448B@kaoexc4.kao.dec.com... >h > mmm.. really mind boggling   That it is.    >mD > >>> With 10,000 servers expected to be installed by the end of the	 year..<<<    Now begins the FUD.a   >r* > - 10,000 servers - how are backups done?  G Primarily via the replication, I'd guess:  when you don't have to worrywL about user screw-ups (almost all of this is read-only content with automatedJ update), the need for backups is reduced to covering software screw-ups if your data is safely replicated.   F If you're really worried, you capture the incoming update stream as itH arrives, in a manner that allows you to recover anything lost:  still noG backups to do, just restore if you need it.  But since at worst you can/F always reconstruct the data anyway from the Web (which in any event is0 happening all the time), you may just not worry.  ; > - 10,000 servers - how are OS upgrades and patches done ?   ? Since the servers are mirrored, presumably by rolling upgrades.r  L > - 10,000 servers - interesting security, data integrity (virus checking ?) > challenges  F Again, no external party does any updating, so I'm not sure what those6 challenges might be (save for having a good firewall).  G > - 10,000 servers - but does not want expensive proprietary servers? I J > suspect NorthernLight and AltaVista may use less than 50-100 each. Would beL > interesting math question when one considers expensive computer room space > A/C etc ..  = Legitimate question, but those 1U boxes stack pretty tightly.m  3 > - 10,000 servers - likely all in one datacenter ?    Not:  read the article.e    What if datacenter gets) > taken out by fire? What is DR solution?aI > - 10,000 servers - interesting financial challenges ie. depreciation bysJ > finance depts. is usually done over three year period, but this presents  > some interesting challenges ..  0 Not unless you depreciate them by hand on paper.  G > - 10,000 servers - what is their upgrade plan to new systems 12-18-24r months > from now?m  L Again, rolling upgrade should work just as well as it does for VMS - if/when they decide they need it.l   - bill   >> > To each his own .. >s > :-)b >t
 > Regards, >s >  > Kerry Main > Senior Consultant, > Compaq Canadat > Professional Services  > Voice : 613-592-4660 > FAX   : 819-772-7036 > Email : kerry.main@compaq.comu >a >' >  > -----Original Message-----. > From: Bill Todd [mailto:billtodd@foo.mv.com]& > Sent: Monday, June 26, 2000 11:43 AM > To: Info-VAX@Mvb.Saic.Com B > Subject: Re: Northern Light vs. Google (and the winner is . . .) >  >n > 7 > Rob Young <young_r@eisner.decus.org> wrote in messageh' > news:VZkzGzloNz$H@eisner.decus.org...n >  > ...e >c" > > Follow and read this overview: > >6 > >8 >fL http://www.intel.com/ebusiness/estrategies/casestudies/snapshots/google_p.ht > mi > >dF > > They have many clusters, each apparently duplicated.  Each cluster@ > > contains partitioned data.  The 1000 chickens versus 64 oxen2 > > are winning.  But more about this in a bit :-) >mH > Not bad.  Kind of surprises me that each uniprocessor Linux system can onlyL > handle 22 - 40 GB of data (with 256 MB - 1 GB of RAM), but it sounds as ifJ > they're limited by the choice of 100 Mbit Ethernet as the interconnect - andaL > the cost of anything faster (plus the multi-PCI, perhaps SMP system neededG > to make good use of it) may be enough higher than a vanilla commodityn system > to discourage that route.s > J > With the newer IDE disks supporting disconnectable operation (and GoogleK > already willing to write software, so they could write the IDE drivers tooF > support this), it becomes even more attractive from the storage costB > viewpoint - and it sounds as if the cost of the IDE drives alone constitutesqD > close to 50% of the cost of each box.  Given this partitionable anI > application, it's almost impossible to compete on purchase price with aoI > larger centralized box using SCSI drives (leaving aside the question ofp whyr? > IDE drives are so much less expensive); given that it's not au
 life-critical I > application, the fact that the hardware is less self-checking and might J > return brief garbage on the occasional failure isn't that important (andF > Linux is pretty stable on the OS end); given the box standardizationL > throughout the structure, sounds as if any failure results in an immediate4 > box-swap, and can then be trouble-shot at leisure. >oI > So the only obvious large-box argument would be the spaghetti of cablesyJ > linking the boxes.  Telcos lived with that problem for a very long time, so  > Google may be able to as well. >m > - bill >n > >n > > Rob  > >s >y >m   ------------------------------  % Date: Tue, 27 Jun 2000 18:01:57 -0500r7 From: "David J. Dachtera" <djesys.nospam@earthlink.net> @ Subject: Re: Northern Light vs. Google (and the winner is . . .)- Message-ID: <39593265.EC22C9B4@earthlink.net>    Bill Todd wrote: [snip] >  > Now begins the FUD.T >  [MAJOR snippage]   HOLY F___ING S__T, BILL!   Will you get a grip??!!p  - I think you've been at the keyboard too long!-  G Fishing trip time! Take two weeks, at least! Probably get yourself backhE on center by then. If not, time to get that job on the garbage truck!-6 (Better pay, more holidays, no pagers, better benies!)   -- 0 David J. Dachterar dba DJE Systemso" http://home.earthlink.net/~djesys/  : Unofficial Affordable OpenVMS Home Page and Message Board:+ http://home.earthlink.net/~djesys/vms/soho/c   ------------------------------  % Date: Tue, 27 Jun 2000 19:36:49 -0400a' From: "Bill Todd" <billtodd@foo.mv.com>t@ Subject: Re: Northern Light vs. Google (and the winner is . . .)( Message-ID: <8jbdp0$efp$1@pyrite.mv.net>  4 Main, Kerry <Kerry.Main@compaq.com> wrote in messageD news:910612C07BCAD1119AF40000F86AF0D805284497@kaoexc4.kao.dec.com... > Bill,e >r > >>> Now begins the FUD.<<< >o( > Sigh .. I'll ignore your pleasantries.  L I'd be less inclined to offer them if I had ever seen you subject VMS to theH same critical scrutiny you give to its competitors.  There's a lot to beD said for a balanced viewpoint, at least if you wish to be considered
 objective.   >cL > >>> If you're really worried, you capture the incoming update stream as itD > arrives, in a manner that allows you to recover anything lost: <<< >v1 > For 10,000 servers with individual disk drives?a  I Well, yuh.  It's not as if there were 10,000 *update streams* coming in -tI update activity is far less frenetic than query activity.  So you capture K the few incoming update streams raw at their reception/concentration points2I (before they're redistributed to the appropriate servers) so that you caneI re-apply them if you need to (rather than scout out the information again H with the bots) - if you're concerned enough to worry about *any* form ofB backup, given the replication level, the relatively secure mode ofK operation, and the fact that normal activity heals all losses anyway, givent a bit of time.   >lD > >>>> - 10,000 servers - how are OS upgrades and patches done ?<<<<I > >>> Since the servers are mirrored, presumably by rolling upgrades. <<<m >tG > Now there's a good career for a few people .. start at server#1, whendH > finished on server #10,000 start on server#1. Of course, this does notJ > consider FW updates, or any of the other issues associated with revision andeJ > configuration management ...like driver versions, patches ect especially* > since different server vendors are used.  L Kerry, these aren't general-purpose systems:  they have a very limited rangeF of activity.  If they worked when they were installed, the chances areK they'll continue working, without benefit of patches or updates, until some I hardware piece breaks (or they become sufficiently obsolete that it makes J sense to replace them).  I'd suspect that even the newest ones are runningE exactly the same software the oldest ones are, or at least old enough 3 (compatible) versions that they're *really* stable.r   >rK > >>> > - 10,000 servers - but does not want expensive proprietary servers?2 I8J > suspect NorthernLight and AltaVista may use less than 50-100 each. Would beL > interesting math question when one considers expensive computer room space
 > A/C etc ..<i >tC > >>Legitimate question, but those 1U boxes stack pretty tightly.<<l >iK > Ok, even if the servers were $1k each (2 drives + 128Mb memory(?) + 100Mbw# > NIC+ additional racking charges),   I Not sure what that last was.  My guess for the rest would be under $400 -sL perhaps well under, in the quantities they're buying.  Certainly under $500,D since I could almost go out and get *one* for that price myself:  noF keyboard, monitor, mouse, floppy drive, CD-ROM, extraneous ports, etc.  (  we are still talking about $10M for theI > servers alone and this does not count the network switches and routers.  Out L > of EACH cabinet, there would be 80 cables for NC connections. Of course, aL > good portion of the network would have to be redundant as well, so the keyJ > routers would be duplicated. 10,000 servers means 10,000 100Mb ports, soK > this would mean approx 150x 64 port or 75x128 port switches - Can someoneeK > comment on how much these cost ? Are these costs factored into the 10,000i > server costs?l  K Someone better acquainted with the technology than I am might be willing to K comment on the configuration itself.  It doesn't require general any-to-anye= high-internal-bandwidth crossbar-style connectivity, but moreuD M-front-end-to-N-back-end connectivity, which may improve its costs.   >c9 > >>>>> - 10,000 servers - likely all in one datacenter ?e >t  > >>> Not:  read the article.<<< >-I > Oops. Here you are right. I missed the text on different facilities. It H > would be interesting to find out if these are duplicate facilities ie. each5 > can function on its own, but thats for another day.t > J > >>> Again, rolling upgrade should work just as well as it does for VMS -% > if/when they decide they need it.<<s >wI > Except that with OpenVMS cluster common system disks, upgrades are only  doneL > once and when the other roots are rebooted, every system is now at the newH > version. Big difference when you have to plan 10,000 individual system2 > upgrades - even if using a streamlined approach.  E As I said, I doubt that upgrades are common.  And given the number of 3 systems, it's reasonable to assume that they have aoF production-line-efficiency mechanism for handling them.  It's also notH impossible that they figure they'll usually wait to do software upgrades* until they want to replace the entire box.   >sH > It appears to work for them. However, I suspect they are looking at HW costs.I > only and not the Operations and non-system related costs of maintaining, suchJ > a complex environment. When datacenter space, configuration and revisionK > mgmt, UPS, air conditioning, network, upgrade planning and implementationvH > costs are considered, I suspect the story might be a little different.  K A little, perhaps.  But very possibly not sufficiently different to dictateaI a different approach.  And it sure must have been an attractive one for ar start-up operation.c   - bill   >d > Again - to each his own. >  > :-)G >f
 > Regards, >o > Kerry Main > Senior Consultant, > Compaq Canadai > Professional Servicesn > Voice : 613-592-4660 > FAX   : 819-772-7036 > Email : kerry.main@compaq.come >e >d >  > -----Original Message-----. > From: Bill Todd [mailto:billtodd@foo.mv.com]& > Sent: Tuesday, June 27, 2000 2:11 PM > To: Info-VAX@Mvb.Saic.ComtB > Subject: Re: Northern Light vs. Google (and the winner is . . .) >f >r >s6 > Main, Kerry <Kerry.Main@compaq.com> wrote in messageF > news:910612C07BCAD1119AF40000F86AF0D80528448B@kaoexc4.kao.dec.com... > >s > > mmm.. really mind boggling >v
 > That it is.i >e > > F > > >>> With 10,000 servers expected to be installed by the end of the > year..<<<' >l > Now begins the FUD.e >o > >c, > > - 10,000 servers - how are backups done? >gI > Primarily via the replication, I'd guess:  when you don't have to worryhD > about user screw-ups (almost all of this is read-only content with	 automated L > update), the need for backups is reduced to covering software screw-ups if! > your data is safely replicated.m >tH > If you're really worried, you capture the incoming update stream as itJ > arrives, in a manner that allows you to recover anything lost:  still noI > backups to do, just restore if you need it.  But since at worst you caneH > always reconstruct the data anyway from the Web (which in any event is2 > happening all the time), you may just not worry. >r= > > - 10,000 servers - how are OS upgrades and patches done ?l >iA > Since the servers are mirrored, presumably by rolling upgrades.  >eK > > - 10,000 servers - interesting security, data integrity (virus checkingm ?) > > challenges >rH > Again, no external party does any updating, so I'm not sure what those8 > challenges might be (save for having a good firewall). >>I > > - 10,000 servers - but does not want expensive proprietary servers? IfL > > suspect NorthernLight and AltaVista may use less than 50-100 each. Would > beH > > interesting math question when one considers expensive computer room spacen > > A/C etc .. >s? > Legitimate question, but those 1U boxes stack pretty tightly.e ><5 > > - 10,000 servers - likely all in one datacenter ?o >n > Not:  read the article.r >  >  What if datacenter gets+ > > taken out by fire? What is DR solution?eK > > - 10,000 servers - interesting financial challenges ie. depreciation by,L > > finance depts. is usually done over three year period, but this presents" > > some interesting challenges .. >c2 > Not unless you depreciate them by hand on paper. > I > > - 10,000 servers - what is their upgrade plan to new systems 12-18-24  > months
 > > from now?  > F > Again, rolling upgrade should work just as well as it does for VMS - if/whenn > they decide they need it.u >  > - bill >/ > >  > > To each his own .. > >, > > :-)e > >o > > Regards, > >g > >< > > Kerry Main > > Senior Consultant, > > Compaq CanadaM > > Professional Servicesa > > Voice : 613-592-4660 > > FAX   : 819-772-7036! > > Email : kerry.main@compaq.com  > >d > >o > >n > > -----Original Message-----0 > > From: Bill Todd [mailto:billtodd@foo.mv.com]( > > Sent: Monday, June 26, 2000 11:43 AM > > To: Info-VAX@Mvb.Saic.ComeD > > Subject: Re: Northern Light vs. Google (and the winner is . . .) > >e > >  > > 9 > > Rob Young <young_r@eisner.decus.org> wrote in message ) > > news:VZkzGzloNz$H@eisner.decus.org...m > >n > > ...  > > $ > > > Follow and read this overview: > > >0 > > >v > >s >.L http://www.intel.com/ebusiness/estrategies/casestudies/snapshots/google_p.ht > > mt > > ><H > > > They have many clusters, each apparently duplicated.  Each clusterB > > > contains partitioned data.  The 1000 chickens versus 64 oxen4 > > > are winning.  But more about this in a bit :-) > >rJ > > Not bad.  Kind of surprises me that each uniprocessor Linux system can > onlyK > > handle 22 - 40 GB of data (with 256 MB - 1 GB of RAM), but it sounds aso ifL > > they're limited by the choice of 100 Mbit Ethernet as the interconnect - > and G > > the cost of anything faster (plus the multi-PCI, perhaps SMP system0 neededI > > to make good use of it) may be enough higher than a vanilla commoditya > system > > to discourage that route.h > >eL > > With the newer IDE disks supporting disconnectable operation (and GoogleJ > > already willing to write software, so they could write the IDE drivers toH > > support this), it becomes even more attractive from the storage costD > > viewpoint - and it sounds as if the cost of the IDE drives alone
 > constitutesnF > > close to 50% of the cost of each box.  Given this partitionable anK > > application, it's almost impossible to compete on purchase price with a K > > larger centralized box using SCSI drives (leaving aside the question ofs > why-A > > IDE drives are so much less expensive); given that it's not a  > life-critical K > > application, the fact that the hardware is less self-checking and mightiL > > return brief garbage on the occasional failure isn't that important (andH > > Linux is pretty stable on the OS end); given the box standardizationD > > throughout the structure, sounds as if any failure results in an	 immediatet6 > > box-swap, and can then be trouble-shot at leisure. > >sK > > So the only obvious large-box argument would be the spaghetti of cables L > > linking the boxes.  Telcos lived with that problem for a very long time, > so" > > Google may be able to as well. > >h
 > > - bill > >e > > >n	 > > > Robk > > >r > >d > >o >i >s   ------------------------------  % Date: Tue, 27 Jun 2000 20:03:43 -0500w* From: Keith Brown <kbrown780@usfamily.net>@ Subject: Re: Northern Light vs. Google (and the winner is . . .), Message-ID: <39594EEF.2E8DD124@usfamily.net>   "Main, Kerry" wrote: >  > Bill,o >  > >>> Now begins the FUD.<<< > ( > Sigh .. I'll ignore your pleasantries. > L > >>> If you're really worried, you capture the incoming update stream as itD > arrives, in a manner that allows you to recover anything lost: <<< > 1 > For 10,000 servers with individual disk drives?t > D > >>>> - 10,000 servers - how are OS upgrades and patches done ?<<<<I > >>> Since the servers are mirrored, presumably by rolling upgrades. <<<g > G > Now there's a good career for a few people .. start at server#1, wheneH > finished on server #10,000 start on server#1. Of course, this does notN > consider FW updates, or any of the other issues associated with revision andJ > configuration management ...like driver versions, patches ect especially* > since different server vendors are used. > M > >>> > - 10,000 servers - but does not want expensive proprietary servers? IaM > suspect NorthernLight and AltaVista may use less than 50-100 each. Would beaL > interesting math question when one considers expensive computer room space
 > A/C etc ..<. > C > >>Legitimate question, but those 1U boxes stack pretty tightly.<<s > K > Ok, even if the servers were $1k each (2 drives + 128Mb memory(?) + 100Mb0K > NIC+ additional racking charges), we are still talking about $10M for therM > servers alone and this does not count the network switches and routers. Out-L > of EACH cabinet, there would be 80 cables for NC connections. Of course, aL > good portion of the network would have to be redundant as well, so the keyJ > routers would be duplicated. 10,000 servers means 10,000 100Mb ports, soK > this would mean approx 150x 64 port or 75x128 port switches - Can someone-K > comment on how much these cost ? Are these costs factored into the 10,000, > server costs?  > 9 > >>>>> - 10,000 servers - likely all in one datacenter ?o >   > >>> Not:  read the article.<<< > I > Oops. Here you are right. I missed the text on different facilities. ItNM > would be interesting to find out if these are duplicate facilities ie. each 5 > can function on its own, but thats for another day.s > J > >>> Again, rolling upgrade should work just as well as it does for VMS -% > if/when they decide they need it.<<  > N > Except that with OpenVMS cluster common system disks, upgrades are only doneL > once and when the other roots are rebooted, every system is now at the newH > version. Big difference when you have to plan 10,000 individual system2 > upgrades - even if using a streamlined approach. > N > It appears to work for them. However, I suspect they are looking at HW costsN > only and not the Operations and non-system related costs of maintaining suchJ > a complex environment. When datacenter space, configuration and revisionK > mgmt, UPS, air conditioning, network, upgrade planning and implementationbH > costs are considered, I suspect the story might be a little different. >  > Again - to each his own. >  > :-)s > 
 > Regards, >  > Kerry Main > Senior Consultant, > Compaq Canadan > Professional Servicest > Voice : 613-592-4660 > FAX   : 819-772-7036 > Email : kerry.main@compaq.comG >  > -----Original Message-----. > From: Bill Todd [mailto:billtodd@foo.mv.com]& > Sent: Tuesday, June 27, 2000 2:11 PM > To: Info-VAX@Mvb.Saic.ComvB > Subject: Re: Northern Light vs. Google (and the winner is . . .) > 6 > Main, Kerry <Kerry.Main@compaq.com> wrote in messageF > news:910612C07BCAD1119AF40000F86AF0D80528448B@kaoexc4.kao.dec.com... > >f > > mmm.. really mind boggling > 
 > That it is.t >  > > F > > >>> With 10,000 servers expected to be installed by the end of the > year..<<<t >  > Now begins the FUD.a >  > >e, > > - 10,000 servers - how are backups done? > I > Primarily via the replication, I'd guess:  when you don't have to worrywN > about user screw-ups (almost all of this is read-only content with automatedL > update), the need for backups is reduced to covering software screw-ups if! > your data is safely replicated.m > H > If you're really worried, you capture the incoming update stream as itJ > arrives, in a manner that allows you to recover anything lost:  still noI > backups to do, just restore if you need it.  But since at worst you can9H > always reconstruct the data anyway from the Web (which in any event is2 > happening all the time), you may just not worry. > = > > - 10,000 servers - how are OS upgrades and patches done ?o > A > Since the servers are mirrored, presumably by rolling upgrades.o > N > > - 10,000 servers - interesting security, data integrity (virus checking ?) > > challenges > H > Again, no external party does any updating, so I'm not sure what those8 > challenges might be (save for having a good firewall). > I > > - 10,000 servers - but does not want expensive proprietary servers? IeL > > suspect NorthernLight and AltaVista may use less than 50-100 each. Would > beN > > interesting math question when one considers expensive computer room space > > A/C etc .. > ? > Legitimate question, but those 1U boxes stack pretty tightly.  > 5 > > - 10,000 servers - likely all in one datacenter ?C >  > Not:  read the article.9 >  >  What if datacenter gets+ > > taken out by fire? What is DR solution?iK > > - 10,000 servers - interesting financial challenges ie. depreciation by L > > finance depts. is usually done over three year period, but this presents" > > some interesting challenges .. > 2 > Not unless you depreciate them by hand on paper. > I > > - 10,000 servers - what is their upgrade plan to new systems 12-18-24  > months
 > > from now?  > N > Again, rolling upgrade should work just as well as it does for VMS - if/when > they decide they need it.l >  > - bill >  > >  > > To each his own .. > >r > > :-)* > >e > > Regards, > >  > >p > > Kerry Main > > Senior Consultant, > > Compaq Canada  > > Professional Servicesi > > Voice : 613-592-4660 > > FAX   : 819-772-7036! > > Email : kerry.main@compaq.comi > >  > >e > >o > > -----Original Message-----0 > > From: Bill Todd [mailto:billtodd@foo.mv.com]( > > Sent: Monday, June 26, 2000 11:43 AM > > To: Info-VAX@Mvb.Saic.ComoD > > Subject: Re: Northern Light vs. Google (and the winner is . . .) > >v > >e > >m9 > > Rob Young <young_r@eisner.decus.org> wrote in messagea) > > news:VZkzGzloNz$H@eisner.decus.org...  > >  > > ..., > >e$ > > > Follow and read this overview: > > >? > > >  > >iN > http://www.intel.com/ebusiness/estrategies/casestudies/snapshots/google_p.ht > > m  > > >rH > > > They have many clusters, each apparently duplicated.  Each clusterB > > > contains partitioned data.  The 1000 chickens versus 64 oxen4 > > > are winning.  But more about this in a bit :-) > >nJ > > Not bad.  Kind of surprises me that each uniprocessor Linux system can > onlyN > > handle 22 - 40 GB of data (with 256 MB - 1 GB of RAM), but it sounds as ifL > > they're limited by the choice of 100 Mbit Ethernet as the interconnect - > andnN > > the cost of anything faster (plus the multi-PCI, perhaps SMP system neededI > > to make good use of it) may be enough higher than a vanilla commodityo > system > > to discourage that route.t > > L > > With the newer IDE disks supporting disconnectable operation (and GoogleM > > already willing to write software, so they could write the IDE drivers toeH > > support this), it becomes even more attractive from the storage costD > > viewpoint - and it sounds as if the cost of the IDE drives alone
 > constitutestF > > close to 50% of the cost of each box.  Given this partitionable anK > > application, it's almost impossible to compete on purchase price with a,K > > larger centralized box using SCSI drives (leaving aside the question ofe > whyeA > > IDE drives are so much less expensive); given that it's not a  > life-critical K > > application, the fact that the hardware is less self-checking and mightpL > > return brief garbage on the occasional failure isn't that important (andH > > Linux is pretty stable on the OS end); given the box standardizationN > > throughout the structure, sounds as if any failure results in an immediate6 > > box-swap, and can then be trouble-shot at leisure. > >oK > > So the only obvious large-box argument would be the spaghetti of cablesbL > > linking the boxes.  Telcos lived with that problem for a very long time, > so" > > Google may be able to as well. > > 
 > > - bill > >d > > >e	 > > > Robe > > >e > >0 > >1    @ Good Grief! What about the cost of the army of support people to; keep it running. Running 10,000 of anything would be a reale bitch.     -- o Keith Brown1 kbrown780@usfamily.net   ------------------------------    Date: 28 Jun 2000 00:09:02 -0500* From: young_r@eisner.decus.org (Rob Young)@ Subject: Re: Northern Light vs. Google (and the winner is . . .)+ Message-ID: <htJgyZcGG9g2@eisner.decus.org>-  R In article <8jbdp0$efp$1@pyrite.mv.net>, "Bill Todd" <billtodd@foo.mv.com> writes: > 6 > Main, Kerry <Kerry.Main@compaq.com> wrote in messageF > news:910612C07BCAD1119AF40000F86AF0D805284497@kaoexc4.kao.dec.com... >> Bill, >>  = 	Even though David J. thinks Bill needs a vacation, I believer< 	he is spot on here.  As Bill points out, this is replicated= 	hither and yon, backups are not necessary.. automated servero> 	setup is a must and as mentioned in the Intel eBusiness piece: 	they "Googal-ize" the servers (envisioning plugging it in> 	and answering a few questions in automated scripts, i.e. IP).  9 	My initial criticisms/cynicisms were along the lines of  : 	employee head count (7000 drives, with great MTBF someone? 	is changing out several drives a week, etc.) power consumption @ 	(Environmentally very harmful.  If nothing else , could rile up; 	a few tree huggers and point out how many resources Googler; 	is wasting... ironically this could be one of the greatestl: 	criticisms).  Not to be overlooked is the whole manner of? 	efficiency which points a finger at the power consumption too.   ; 	It seems they want it both ways, one argument for multiplee+ 	PCs is because of computational intensity:e  2 http://www.techweb.com/wire/story/TWB20000530S0011  ; 	Yeah, right.  Bet those 4000 CPUs are really humming doingt@ 	13 million searches per day... at 8 million searches peak hoursA 	i.e. 8 hour period, that's 278 searches per second.  Yeah, I bet & 	those 4000 CPUs are really straining.  8 	But more realistically, from the Intel eBusiness piece:  M http://www.intel.com/ebusiness/estrategies/casestudies/snapshots/google_p.htmn  H "In Google's environment, disk I/O performance is an overriding factor	"  : 	In fairness, maybe each PC has data and index such that aA 	portion of the index is cachable (i.e. hot)... unscientific test-< 	of "monica clinton" takes .24 seconds and a retry takes .03@ 	seconds.  Smart enough to redirect to same cluster and the term? 	is still in cache at 10:35 p.m. EDT.  Again, "hillary chelsey" = 	takes .27 seconds, retry .05 seconds.  A term that is likely = 	in cache?  How about genome?  Yep.  .06 seconds.  Now, let'srC 	see how "genome soup" does .. .37 seconds.  Terms to be in memory:F; 	"genome dna" , so it does do quite well on cached index.  y  ? 	My point is that not all index is cached and when it has to goe: 	to disk, real-world results.  Again, a single term not in? 	cache:  "zugswang" took .31 seconds.  No "and" involved so theh? 	overhead of .06 seconds is saved.. thankfully not taxing thosee= 	very busy CPUs.  But that's a good data point.. depending onoC 	the term it takes .24 to .31 seconds (if not in cache I suppose)..n! 	i.e. 240 to 310 milliseconds ;-)i  B 	But not much can be cached apparently as 3 minutes later "clintonD 	soup" takes .54 seconds to resolve [oops, not true, "soup" isn't in5 	cache, see below].  Hillary is still there, "clintoni@ 	hillary" takes .06 seconds.  "dna soup" takes .49 seconds.  Ah,A 	soup is finally cached on "genome soup" so perhaps it takes 5 orfD 	6 go rounds to trick up the caching algorithm.  Yep, "chelsey soup"E 	takes a while (chelsey isn't cached) but "dna soup" is now flying.  tE 	How about "doj"?   Yep.. .06, "reno"?  .31 seconds.  "clinton reno"?nB 	Slow...  Footnote:  Took about 10 go rounds to force chelsey into> 	cache.. so maybe someone else had searched for "soup" lately.   	Enough.    ? 	For now trying to come up with a rebuttal to Bill's valid casebA 	in his initial reply.  Didn't want to be too flippant and I wantt@ 	to slightly tweek the initial scenario.  Still believe Northern> 	Light will come out on top in all this but it will take about8 	a year or a year and a half for Marvel (Terabyte Memory 	machine) to arrive (guess).  > 	Best campaign Northern Light could come up with then would beE 	the real-time indexing aspect and a shot of their parking lot and a eE 	shot of their lights-out operation with the staff reading magazines  G 	and cut to a shot to a nameless Internet Search Service abuser of the sB 	environment showing shots of their parking lots and the acres of F 	servers and a quick pan to the power meters out back that have smoke B 	coming off of them because they are whizzing around so fast.  To G 	really stick it to them, perhaps a few fish floating on the surface ofs: 	a pristine stream; boiled alive by nuclear heat effluent.  % 	More about real-time indexing later.    				RobS   ------------------------------  % Date: Wed, 28 Jun 2000 00:45:47 -0400e- From: JF Mezei <jfmezei.spamnot@videotron.ca>d@ Subject: Re: Northern Light vs. Google (and the winner is . . .), Message-ID: <395982F0.AD142DC1@videotron.ca>   Keith Brown wrote:B > Good Grief! What about the cost of the army of support people to= > keep it running. Running 10,000 of anything would be a reali > bitch.  G If this is funded by Intel, then it doesn't matter how much it costs toMJ maintain or how difficult such a 10000 billybox network is to manage. WhatJ matters is that Intel can claim its 8086 is fully capable to process large+ amounts of data just like the big guys can.o  L Think of Northen Light which was somewhat a poster boy for VMS, proving thatM VMS could be a "me too" in the search engine business. So Intel also wanted ag "me too" and now they have one.   H And if they can show that they have mamangement tools to manage 1000s ofJ billyboxes, then it becomes great advertising to companies who may have toI manage 100s of billyboxes since they are confident that this software can  handle their much smaller load.y   ------------------------------    Date: 28 Jun 2000 02:49:16 -0500* From: young_r@eisner.decus.org (Rob Young)@ Subject: Re: Northern Light vs. Google (and the winner is . . .)+ Message-ID: <sqOSMwTvVkag@eisner.decus.org>h  \ In article <395982F0.AD142DC1@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> writes: > Keith Brown wrote:C >> Good Grief! What about the cost of the army of support people toE> >> keep it running. Running 10,000 of anything would be a real	 >> bitch.  > I > If this is funded by Intel, then it doesn't matter how much it costs toiL > maintain or how difficult such a 10000 billybox network is to manage. WhatL > matters is that Intel can claim its 8086 is fully capable to process large- > amounts of data just like the big guys can.a > N > Think of Northen Light which was somewhat a poster boy for VMS, proving thatO > VMS could be a "me too" in the search engine business. So Intel also wanted a ! > "me too" and now they have one.s > J > And if they can show that they have mamangement tools to manage 1000s ofL > billyboxes, then it becomes great advertising to companies who may have toK > manage 100s of billyboxes since they are confident that this software canb! > handle their much smaller load.a    A 	Ummm.. couple things here.  www.alltheweb.com is powered by DellsA 	servers, Intel is here and there.  Also, the Google servers are a/ 	running Linux.  "Billyboxes" isn't a good fit.o   Asides (building a case :-):  F 	You can get most anything to scale.  In fact as Bill Todd pointed out; 	recently, you can partition any database.  It may not make ; 	a whole lot of sense to partition a customer database into ; 	20 parts but if you had a million customers it just might.   = 	However, if you had massive enough resources that could more F 	than handle the 1 million customers as a single database (all things B 	considered, search results, index times, backups, etc.)... it may? 	make little sense to break it up from a managerial standpoint n+ 	(and other equally debatable points :-).  e 	y? 	Found an article that points out Google is re-indexed monthly.,? 	Alltheweb claims to be able to reindex in 12  hours.  And yet, C 	it is quite apparent that alltheweb hasn't been reindexed recently  	as:  + lakers champions basketball nba pacers riot   G 	turns up nothing relevant and the same search on www.northernlight.com F 	(Web only search) gets a few relevant hits (and a few junk hits too).  C 	Weirdly, Google has a single hit.  Saw this with another search onoG 	Google.  Google is apparently re-indexing select sites maybe as often   	as weekly.    				Robt   ------------------------------  # Date: Wed, 28 Jun 2000 04:08:59 GMTi0 From: Timothy Stark <sword7@grace.speakeasy.org> Subject: ODS2 Reader/Writera8 Message-ID: <vTe65.282225$VR.4160411@news5.giganews.com>   Hello Folks:  J I am looking for the ODS2 reader/writer that allow transfer files from andD to ODS2-based disk image files or CD-RW, etc. on other platform like5 Windows 98, Linux, etc.  Does anyone know about that?t  
 Thank you!   -- Tim Stark   --  C Timothy Stark	<><	Inet: sword7@speakeasy.org, sword7@firesword7.net J --------------------------------------------------------------------------F "For God so loved the world, that he gave his only begotten Son, that H whosoever believeth in him should not perish, but have everlasting life.. Amen." -- John 3:16 (King James Version Bible)   ------------------------------   Date: 27 Jun 2000 20:39:24 GMT2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)1 Subject: Re: OpenVMS Alpha C programming questiono, Message-ID: <8jb3ds$r2n@gap.cco.caltech.edu>  ^ In article <39578C0B.1F1A9012@hydrogen.ucsc.edu>, Matt Evans <evans@hydrogen.ucsc.edu> writes:F >I have just begun experimentation with 64-bit unsigned long long dataE >types.  I am normally a UNIX programmer and I am somewhat unfamiliar G >with the the OpenVMS environment (specifically, 64-bit C programming).cB >My question (and statement) is that I can't seem to get the shiftH >operators to work, i.e. 1<<63 does not give a correct result, it simplyI >gives zero.  The shift operator only seems to work with shifts less thanrF >32 bits.  However, I can create a bitmask (example 0x8000000000000000@ >which is the hexadecimal equivalent of the 64th bit) and use itD >successfully.  I just can't seem to get the shift operators to work= >corectly with the 64 bit unsigned long longs.  Please help!?o  F Please supply a small program which demonstrates the problem.  Include4 the compiler and link commands you used to build it.   Regards,   David Mathog mathog@seqaxp.bio.caltech.edue? Manager, sequence analysis facility, biology division, Caltech o   ------------------------------   Date: 27 Jun 2000 22:35 PST ) From: rankin@eql.caltech.edu (Pat Rankin)r1 Subject: Re: OpenVMS Alpha C programming question / Message-ID: <27JUN200022354046@eql.caltech.edu>n  2 In article <39578C0B.1F1A9012@hydrogen.ucsc.edu>,\/  Matt Evans <evans@hydrogen.ucsc.edu> writes... G > I have just begun experimentation with 64-bit unsigned long long dataTF > types.  I am normally a UNIX programmer and I am somewhat unfamiliarH > with the the OpenVMS environment (specifically, 64-bit C programming).C > My question (and statement) is that I can't seem to get the shiftrI > operators to work, i.e. 1<<63 does not give a correct result, it simply.J > gives zero.  The shift operator only seems to work with shifts less thanG > 32 bits.  However, I can create a bitmask (example 0x8000000000000000eA > which is the hexadecimal equivalent of the 64th bit) and use itsE > successfully.  I just can't seem to get the shift operators to workd> > corectly with the 64 bit unsigned long longs.  Please help!?  @      The constant `1' has type `int', and int is only 32 bits in? the Alpha/VMS C implementation.  When I try to use ``1 << 63'', 9 the compiler gives me a moderately clear warning message:w  H %CC-W-SHIFTCOUNT, In this statement, the shift count "63" is negative orB  is greater than or equal to the promoted size of the operand "1".  ?      When I use ``1ULL << 63'', I get the result you seem to bea> expecting. Explicitly specifying `unsigned long long' yields a- value that can be shifted by 32 or more bits.   2                 Pat Rankin, rankin@eql.caltech.edu   ------------------------------  % Date: Tue, 27 Jun 2000 14:52:49 -0400o From: jlahman@LTVSteel.com Subject: OT - FEELING OLD!!s8 Message-ID: <8525690B.00686B7B.00@notesnta.LTVSteel.com>  > --0__=pD9NHwicCrEYtMff7y55Um5qOhmvHmNWaC98fQICMDh37vH9E8ty6WAp* Content-type: text/plain; charset=us-ascii Content-Disposition: inline     N ---------------------- Forwarded by Jim Lahman/CLWK/LTV on 06/27/2000 02:46 PM ---------------------------iG                                                                        tG                                                                         G  (Embedded     Dawn Yonek                                               G  image moved   06/27/2000 02:40 PM                                     aG  to file:      (Embedded image moved to file: pic13488.pcx)            DG  pic19672.pcx)                                                          G                                                                        o      K To:   Diana Conway/CLWK/LTV@LTVProd, Mark Miller/CLWK/LTV@LTVProd, ShailesheB       Thakkar/CLWK/LTV@LTVProd, Jim Lahman/CLWK/LTV@LTVProd, DorelA       Stefan/CLWK/LTV@LTVProd, Yilmaz Rona/CLWK/LTV@LTVProd, Bill E       Aiello/CLWK/LTV@LTVProd, John Novotny/CLWK/LTV@LTVProd, RichardSA       Vincent/CLWK/LTV@LTVProd, John Zazik/CLWK/LTV@LTVProd, Jack C       Burns/CLWK/LTV@LTVProd, Michael Kovatch/CLWK/LTV@LTVProd, Bob >       Bing/CLWK/LTV@LTVProd, Keith Ross/CLWK/LTV@LTVProd, John       Zilka/CLWK/LTV@LTVProd cc:  Subject:  FEELING OLD!!     / Just in case you weren't feeling too old today,  > > > this will certainlyi5 > > > > >change things. Each year the staff at Beloitf > > > College in Wisconsin putse5 > > > > >together a list to try to give the faculty a " > > > sense of the mindset of that7 > > > > >years incoming freshmen.   Here is this year'so
 > > > list	 > > > > >u6 > > > > >The people who are starting college this fall! > > > across the nation were born  > > > > >ina > > > > >1982.	 > > > > > 4 > > > > >They have no meaningful recollection of the! > > > Reagan Era and probably didl	 > > > not $ > > > > >know he had ever been shot.9 > > > > >They were prepubescent when the Persian Gulf Wari > > > was wagedt7 > > > > >Black Monday 1987 is as significant to them as  > > > the Great Depression.:& > > > > >There has been only one Pope.7 > > > > >They were 11 when the Soviet Union broke apart  > > > and do not remember the 
 > > > Cold > > > > >War. > > > > >They have never feared a nuclear war.9 > > > > >They are too young to remember the space shuttle  > > > blowing up.g. > > > > >Tianamen Square means nothing to them3 > > > > >Bottle caps have always been screw off andh > > > plastic.5 > > > > >Atari predates them, as do vinyl albums. The ! > > > expression you sound like as- > > > > >broken record means nothing to them. 5 > > > > >They have likely never owned a record playeri7 > > > > >They have likely never played Pac Man and haven > > > never heard of Ping-Pong1 > > > > >They may have never heard of an 8 track.t9 > > > > >The Compact Disc was introduced when they were 1o > > > year old.h5 > > > > >As far as they know, stamps have always cost' > > > about 33 cents2 > > > > >They have always had an answering machine3 > > > > >Most have never seen a TV set with only 13i$ > > > channels, nor have they seen a > > > > >black-and-white TVh# > > > > >They have always had cable : > > > > >There has always been VCRs, but they have no idea > > > what BETA is.u7 > > > > >They cannot fathom not having a remote controlo2 > > > > >They were born the year that Walkmen were > > > introduced by Sony.a8 > > > > >Roller-skating has always meant inline for them6 > > > > >Jay Leno has always been on the Tonight Show.: > > > > >They have no idea when or why Jordache jeans were > > > cool.e9 > > > > >Popcorn has always been cooked in the microwave... > > > > >They have never seen Larry Bird play.6 > > > > >They never took a swim and thought about Jaws9 > > > > >The Vietnam War is as ancient history to them as  > > > WWI, WWII and the Civils
 > > > > >War.f8 > > > > >They have no idea that Americans were ever held > > > hostage in Iran.9 > > > > >They can't imagine what hard contact lenses are.-5 > > > > >They don't know who Mork was or where he was: > > > from.m: > > > > >They never heard: Where's the beef?, I'd walked a > > > mile for a Camel,h! > > > > >or>de>>> plane, de planeJ8 > > > > >They do not care who shot J.R. and have no idea > > > who J.R. is.6 > > > > >The Titanic was found? They thought we always > > > knew where it was./ > > > > >Michael Jackson has always been white.6: > > > > >Kansas, Chicago, Boston, America, and Alabama are > > > places, not groups6 > > > > >McDonalds never came in Styrofoam containers." > > > > >There has always been MTV7 > > > > >They don't have a clue how to use a typewriters7 > > > > >Do you feel old yet? pass this on to the otheru > > > old fogies	 > > > > >z > > >s >e >( >i            > --0__=pD9NHwicCrEYtMff7y55Um5qOhmvHmNWaC98fQICMDh37vH9E8ty6WAp( Content-type: application/octet-stream;  	name="pic19672.pcx"8 Content-Disposition: attachment; filename="pic19672.pcx"! Content-transfer-encoding: base64l  L CgUBCAAAAAArABoAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAABLAABAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAADID9MAyQAAxA/CDw/JD9IAyQDFD8IPD8oP0gDHAMUPww8Pyw/RAMYAxg/DL Dw/MD9EAxADGD8MPwg/NDdAAwwDHDcMNwg3OD9AAAMcPxA/CD88PzwDID8QPwg/DD80NzQDIDcQNL wg8P0Q/LAMkPxA/CDw/SD8kAyQ/FD8IPD8YPzQ3HAMoNwg3ED8IP1A/FAMoPxQ/DDw/VD8MAyw/FL D8MPD8kPzQ0Ayw0NxQ/DDw/XD8sPxg/DDw/XD8sPxg/DDw/MD9ENww3HD8MPwg/XD8sPxg/DDw/XL D8sPxg/DDw/PD84NyA/ED8IPD9cPyw/GD8MPD9cPyw/GD8MPD9IPyA3KD8UPwg8P1w/LD8YPww8PL 1w/LD8YPww8P1Q/CDcsPxg/DDw8MAAAAgAAAAIAAgIAAAACAgACAAICAgICAwMDA/wAAAP8A//8AL AAD//wD/AP//////AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA@ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA  > --0__=pD9NHwicCrEYtMff7y55Um5qOhmvHmNWaC98fQICMDh37vH9E8ty6WAp( Content-type: application/octet-stream;  	name="pic13488.pcx"8 Content-Disposition: attachment; filename="pic13488.pcx"! Content-transfer-encoding: base64a  L CgUBCAAAAABSAQkAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAABUwEBAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAD/D/8P/w//D+wP1g/LD8YPww8Pwg/DBcgPwwXID8MFyA/DBcgPwwXID8MFL yA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXIL D8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcMPDw/FBcYPxQXGD8UFL xg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGL D8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcIPL Dw/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGL D8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPL xQXGD8UFxg/FBcIPDw/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGL D8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPxQXGD8UFxg/FBcYPL xQXGD8UFxg/FBcYPxQXGD8UFxg/FBcIPD8IPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFL yA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXIL D8MFyA/DBcgPwwXID8MFyA/DBcgPwwXID8MFyA/DBcgPwwXDDw//D/8P/w//D+wP1g/LD8YPww8PL /w//D/8P/w/sD9YPyw/GD8MPD/8P/w//D/8P7A/WD8sPxg/DDw//D/8P/w//D+wP1g/LD8YPww8PL DAAAAIAAAACAAICAAAAAgIAAgACAgMDAwICAgP8AAAD/AP//AAAA//8A/wD//////wAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAL AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA( AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA==  @ --0__=pD9NHwicCrEYtMff7y55Um5qOhmvHmNWaC98fQICMDh37vH9E8ty6WAp--   ------------------------------   Date: 27 Jun 2000 20:13:45 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: OVMS 7.2 updates 6 Message-ID: <8jb1tp$bkm$3@mailint03.im.hou.compaq.com>  ^ In article <fkfhls885kpjgggncf6rc85nked59lm102@4ax.com>, Tom Dockray <dockray@acm.org> writes:C :I am running OVMS 7.2 on a MV3100 under the hobbiest license. Just.E :wondering if it is advisable to apply the updates I stumbled upon oneG :the Compaq site. Does anybody have any experiences with this that theys :would like to share?   K   Um, which updates are you refering to?  (Sorry, I left my psychic powers a(   on the kitchen table this morning. :-)  F   Are you seeing any particular problems with your OpenVMS VAX system?  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Tue, 27 Jun 2000 17:56:28 -0500e7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>r Subject: Re: print to file- Message-ID: <3959311C.3648C21D@earthlink.net>    Bob Ricci wrote: > M > Can anyone help me print to a file- I will be more specific !We really wantIJ > to email a .rpt  file from open alpha vms 6.1 to a pc but when you go to$ > print it, it seems to wrap around.  1 Hhmmm... sounds like an improper transfer method.   F Remember that "report" files on OpenVMS are usually plain ASCII "text"@ files. That means they must be transferred as ASCII, not binary.  * > The consensus here is if I could somehow< > issue a print statement with an output qualifier such as aI > dir/output=file.rpt it can then be played with , sent to a pc and printhI > correctly. In other words the print command puts the correct escape and D > control characters necessary to format and print this file on vms.  A Well, there are no "escape characters" except those specific to arH make/model of printer, and most report programs don't handle that - that; comes later, and its usually handled by the print symbiont.   H Unless the report files are generated as STREAM format files (Stream_CR,B Stream_LF or just plain Stream), they usually contain no "control"C characters (those whose ASCII value is less than 32 or greater thany 126).s  
 > We wouldA > like to get this report out to the field as easily as possible.   1 Try transferring it as ASCII. That should do it. t  B Otherwise, be prepared to supply even MORE detail about what it isG you're trying do. Actually, it's best to do that anyway, to help ensuregC that you'll get the answer you need instead of a lot of generalizedy; speculation while respondents try to "cover all the bases".i   -- ( David J. Dachterae dba DJE Systems)" http://home.earthlink.net/~djesys/  : Unofficial Affordable OpenVMS Home Page and Message Board:+ http://home.earthlink.net/~djesys/vms/soho/i   ------------------------------  % Date: Wed, 28 Jun 2000 12:02:28 +1200 - From: Raji Arulambalam <rajia@envbop.govt.nz>  Subject: SMTP Mail@ Message-ID: <F15BB07EE56ED211855600A0C913B9313E84F5@apollo.ebop>   Hi  C I've just installed OVMS 7.2 on a DEC 2000/300 AXP with TCPIP v 5.0e  L I am trying to get SMTP mail working, but when I try to use VMS mail and tryI to use this format   smtp%"rajia@envbop.govt.nz"  the mail program hangs.  Is there some thing missing.??  6 Used to work in v 6.1-2 fine on this machine with UCX.   Any help would be appreciated.   TIAO  - ---------------------------------------------4   Raji Arulambalam           Systems Administrator      f!   Bay of Plenty REGIONAL Council w   P O Box 364 Whakatane.   NEW ZEALAND  k3   Phone: 0800 ENV BOP (0800 368 267) +64 7 921 3390i4   Fax:    0800 ENV FAX (0800 368 329) +64 7 921 3393   http://envbop.govt.nza3 The Rules Have Changed...Get Paid to Surf the Web! o0 http://www.alladvantage.com/go.asp?refid=LJG413 , --------------------------------------------H "We were investigating... that's all we ever do around here. Why pretendI we're going home at all? All we're going to do is investigate every cubicf@ millimeter of this quadrant, aren't we?" The Holodoc (The Cloud)      6 ******************************************************F This e-mail has been checked for viruses and no viruses were detected.   ------------------------------   Date: 28 Jun 2000 04:58:21 GMT) From: leslie@clio.rice.edu (Jerry Leslie)u Subject: Re: SMTP Mail' Message-ID: <8jc0ld$1ie$1@joe.rice.edu>u  . Raji Arulambalam (rajia@envbop.govt.nz) wrote: : Hi  E : I've just installed OVMS 7.2 on a DEC 2000/300 AXP with TCPIP v 5.0a  J : I am trying to get SMTP mail working, but when I try to use VMS mail andH : try to use this format   smtp%"rajia@envbop.govt.nz"  the mail program' : hangs. Is there some thing missing.??c  8 : Used to work in v 6.1-2 fine on this machine with UCX.    : Any help would be appreciated.   : TIA   B The "smtp%" is no longer necessary as of VMS 6.2, per the VMS FAQ:  :     http://www.openvms.digital.com/wizard/openvms_faq.html  L  "MAIL2.  How do I get IN% or MX% added automatically to Internet addresses?   .h   .6   . J   As of OpenVMS V6.2, this is not necessary -- simply enter the SMTP email   address directly."    4 --Jerry Leslie     (my opinions are strictly my own)   ------------------------------  % Date: Wed, 28 Jun 2000 17:35:12 +1200e- From: Raji Arulambalam <rajia@envbop.govt.nz>, Subject: SMTP Mail - followup-@ Message-ID: <F15BB07EE56ED211855600A0C913B9313E84F6@apollo.ebop>   Hi  9 Installed OVMS 7.2 on a DEC 2000/300 AXP with TCPIP v 5.0e  K Further to my earlier request for help, I have found that the smtp receiverBL on this machine is not responding to connect requests from other machines orG from VMS Mail. ( telnet vaxmail /port=25 works from another DEC machinet8 running UCX 4.0, but hangs when coming from another OS).   Is there a fix.??i   TIAt  - ---------------------------------------------o   Raji Arulambalam       a   Systems Administrator      e!   Bay of Plenty REGIONAL Council o   P O Box 364 Whakatane.   NEW ZEALAND  e3   Phone: 0800 ENV BOP (0800 368 267) +64 7 921 3390n4   Fax:    0800 ENV FAX (0800 368 329) +64 7 921 3393   http://envbop.govt.nzm3 The Rules Have Changed...Get Paid to Surf the Web! d0 http://www.alladvantage.com/go.asp?refid=LJG413 , --------------------------------------------: If a turtle doesn't have a shell, is he homeless or naked?      6 ******************************************************F This e-mail has been checked for viruses and no viruses were detected.   ------------------------------  % Date: Tue, 27 Jun 2000 23:03:27 -0400g2 From: "Richard B. Gilbert" <DRAGON@compuserve.com>* Subject: VAX based 1032 database to Oracle7 Message-ID: <200006272303_MC2-AA54-BBEB@compuserve.com>.  %         VAX is a family of computers.   J         VMS is an operating system that runs on both the VAX architecture=  F and the Alpha architecture.  VAX is CISC while Alpha is RISC.  The VAXG architecture is pretty much obsolete.  The Alpha architecure offers the-J fastest iron that money will buy!  VMS is designed for high availabilty a= nd reliability.  C         1032 is a database.  It was once offered by CompuServe Data_J Technologies but I believe it has since been sold; I don't recall to whom= .p  H         Oracle is a database.  It runs on VAX/VMS, Alpha/VMS,  Sun/Unix,9 Windows, and quite likely, platforms I've never heard of.M  J         The client wants to go to a modern architecture; e.g. Alpha, and = a 6 mainstream data base, which Oracle is and 1032 is not.  C         If I may say so, you, and your company, seem singularly ill H prepared to offer this customer what he wants;  you don't even speak his	 language.   B         I've no idea what a "System Monitoring database" might be.    4 Message text written by INTERNET:mschaus@HOTMAIL.COMD >My company is currently working on a proposal for a project dealingE with VAX systems.  First some simple questions.  Is VAX the hardware,nB operating system, or what?  And how does VMS fit in?  What type of@ databases (Oracle, Access, SQL, etc.) can be run on the systems?  G Specifically, the client wants to "reprogram the current VAX based 1032 G database into an Oracle based platform."  Does anyone have any ideas asi@ to what the exact meaning of this is?  Is 1032 a hardware model,F database type, or maybe a proprietary name?  1032 is used as a "SystemG Monitoring (SYSMON) database," but this has little meaning to me, as itL% could be a product or a made-up name.!  & Any help would be greatly appreciated. <o   ------------------------------  # Date: Wed, 28 Jun 2000 04:25:11 GMT & From: Dave Stone <dave.stone@mail.com>. Subject: Re: VAX based 1032 database to Oracle( Message-ID: <39597E62.81A7CB0C@mail.com>   Larry Kilgallen wrote:  9 > I am not sure what your definition of "network" is, but 5 > my recollection is that 1032 is neither Codasyl nore7 > Relational.  (You seem to apply the term "network" toi= > both 1032 and Oracle DBMS, and they bear little resemblancel > to each other.)e >   b As the person responsible for the development of System 1032, I'd like to set straight some of the+ information posted in this thread recently.   j System 1032 is best described as a Drill-Down engine with relational overtones. Unlike most other databasek systems, 1032 maintains a context of previous searches within the tables, i.e. a subset of records matchingmj the search criterea. Subsequent searches can be constrained to within the existing context and generate ang even smaller subset. Datasets (tables to the SQL minded folk out there) can be joined and the resulting k superset of records searched in the same manner. This methodology closely follows how people typically mineyj for data (as may be observed from how linked web pages and web searches are organized and how OLAP systems are mined).k  h One thing System 1032 is not is it's not an SQL based OLTP RDBMS. The language used is a proprietary 4GLh called PL1032 (sort of a cross between BASIC and Pascal with the data referencing commands thrown in forj good measure). There's no comparison with SQL at all and is, in my biased opinion, much easier to use thank most other databases. We're certainly not CODASYL compliant, and are definitely not a network database. ForyW those customers who want an SQL front-end there is an ODBC driver available for the PC.a  j Finally, to bring to close questions brought up earlier in this thread, yes it runs on the Alpha and underf current versions of OpenVMS, it is supported and undergoing development, and no, I have no idea what a SYSMON database is either.  
 Dave Stone  Manager, System 1032 Development Computer Corporation of Americal  > Standard disclaimers apply (My opinions are my own, etc. etc.)   ------------------------------  # Date: Wed, 28 Jun 2000 01:56:06 GMTc2 From: "Zane H. Healy" <healyzh@shell1.aracnet.com> Subject: Re: VAX on Intel?4 Message-ID: <WWc65.311$9e.67363@typhoon.aracnet.com>   hobbsb@my-deja.com wrote:cE > Compaq gives $500,000 or so to a university to develop the card andfJ > software.  The requirements are that the card will run VMS or Linux in aJ > commodity desktop Wintel or Macintosh box.  The host runs a program thatH > acts like the console and presents the host's devices and ports to VMS7 > or Linux.  VMS and Linux get some new device drivers.e  D Gahck!!!  I'd love something like this for my G4/450 if it would runK OpenVMS, especially since I'm already using eXodus running on it to connectc to my PWS433au!t  B > Grab some Linux geeks.  Ask them if they would like to run theirJ > favorite OS on an Alpha chip?  On their current Wintel or Macintosh box?H > Would they pay $500 to $1000 to do this?  Line forms over there - have > your credit card ready.e  K It would have to cost more than $1000 I suspect.  After all, what's the CPUsJ itself cost, not even counting the board.  But make it take standard PC100I or PC133 DIMMS, and have at least two DIMM sockets, as well as a built inqL U3W-SCSI interface and you'd have a winner!  My guess is the board could be L done for somewhere between $1000-1500.  If it ran OpenVMS and worked with a  Mac, I'd buy one.c  ? But this is only dreaming, I don't ever see this happening. :^(A   			ZaneA   ------------------------------  % Date: Tue, 27 Jun 2000 20:05:14 +0200A2 From: martin@radiogaga.harz.de (Martin Vorlaender)D Subject: Re: what layered products can i install with these licences; Message-ID: <3958ecda.524144494f47414741@radiogaga.harz.de>M  6 David J. Dachtera (djesys.nospam@earthlink.net) wrote:E : NAS-150 (I believe) gets you DECnet end-node (IV or V), DECwindows,g( : MOTIF and UCX (but, I could be wrong).  C IIRC, it's all of UCX except for the NFS server - something like anP? UCX client, but supporting Telnet and FTP server functionality./   cu,/   Martin --J One OS to rule them all       | Martin Vorlaender  |  VMS & WNT programmer7 One OS to find them           | work: mv@pdv-systeme.deAN One OS to bring them all      |       http://www.pdv-systeme.de/users/martinv/> And in the Darkness bind them.| home: martin@radiogaga.harz.de   ------------------------------   Date: 27 Jun 2000 23:37:52 GMT) From: leslie@clio.rice.edu (Jerry Leslie)AF Subject: Will OpenVMS/VAX 7.3 Include VMSINSTAL Installation History ?' Message-ID: <8jbdsg$cbp$1@joe.rice.edu>A0 Keywords: vax,vms,vmsinstal,installation,history  = In a posting to the vmsnet.sysmgt newsgroup on March 2, 1998,A4 message-ID 6df3qj$ubf@usenet.pa.dec.com, Hoff wrote:  @   "We have also extended OpenVMS VAX VMSINSTAL to keep a history@    of product installations in a log, such as is already done on    OpenVMS Alpha."  A This enhancement didn't make it into VMS 7.2. Will it be in 7.3 ?A    4 --Jerry Leslie     (my opinions are strictly my own)   ------------------------------   End of INFO-VAX 2000.358 ************************type: application/octet-stream;  	name="pic13488.pcx"8 Content-Disposition: attachment; filename="pic13488.pcx"! Content-transfer-encoding: base64a  L CgUBCAAAAABSAQkAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA good thing, and 3) I need to ask more/ > questions as to the setup of the system disk.V > F > What I Plan to Do:  as many physical disk drives as possible will beG > put in a single Raid5 set, with this set being broken up into variousuG > logical drives as needed by the applications.  This would spread datatF > access across as many physical spindles as possible and minimize theF > storage loss due to Raid5.  A hot spare would be configured as well.G > As to the number of physical drives in a Raid5 set,  I have not heardiE > that there was a point of diminishing return, so if there is one, IsH > want to know!   The sales rep has suggested that we use a 9 gig systemG > disk using the ES40's "onboard" SCSI controller.  Dual disks would be G > supplied, so shadowing would be possible.  I would prefer to keep thesF > system disk as part of the raid configuration, and not have internalC > disks at all.  I am interested in knowing what others use for thes1 > system disk, given similiar i/o configurations.p > F > To answer some other questions:  this is a single ES40 with dual 677I > mhz processors; a DS-TZ89N-VW DLT drive will be used for backup; expectlJ > approximately 400 concurrent telnet sessions to SCT's Plus2000 software. > - > Thanks again.  All replies are appreciated.  >  > Stanley Hipplern > McNeese State University > + > In article <8glrmu$2i4$1@nnrp1.deja.com>,-1 >   Stanley Hippler <shippler@my-deja.com> wrote:oJ > > From our local sales rep, we requested an ES40 configuration; dual 667E > > pr