1 INFO-VAX	Tue, 25 Apr 2000	Volume 2000 : Issue 230       Contents:3 Re: "Reverse" Telnet as AlphaServer serial console? 3 Re: "Reverse" Telnet as AlphaServer serial console? ! Re: Another VAX to Alpha question  Awesome Opportunity!  264 0 Re: Catastrophic EVE/TPU documentation error :-)" Fall 1999 VMS/LT Sigtapes released ftp.decus.org progress?  Re: ftp.decus.org progress?  Re: ftp.decus.org progress?  Re: ftp.decus.org progress? ! Re: Help needed by non-VMS person 2 re: how do i kwno how much memory is on my system?2 Re: how do i kwno how much memory is on my system?2 Re: how do i kwno how much memory is on my system?& Re: LK46W-A2 & "DO" key in Reflections
 No Subject Re: No Subject Re: NT Breaks Again  Re: OpenVMS marketing  Re: OpenVMS marketing  Re: OpenVMS marketing / Problem with upgrading an Alpha Server to 7.2-1  Recent SKC Postings # Re: system disk move and batch jobs # Re: system disk move and batch jobs # Re: system disk move and batch jobs & Re: The Movie "Breaking Point" + PDP11 Re: Using DiskMizer with Oracle * Re: VAX CI storage vs served fibre channel* Re: VAX CI storage vs served fibre channel RE: Verify of Backups   F ----------------------------------------------------------------------   Date: 24 Apr 2000 17:21:49 CST; From: wayne@tachyon.xxx.394981.killspam.03a1 (Wayne Sewell) < Subject: Re: "Reverse" Telnet as AlphaServer serial console?. Message-ID: <z9+d55648QT7@tachxxsoftxxconsult>  ^ In article <8e1j6q$ar$1@news.enteract.com>, "Dale A. Dellutri" <ddellutr@enteract.com> writes:@ > I will soon have 7 separate ES40s, each of which need a system@ > console.  When I just 3, I used a VT520 with three sessions to@ > have three serial consoles with one screen.  Now, with 7, that > won't be practical.  > @ > I'm considering doing some sort of "reverse" telnet setup: useA > a simple terminal server with 8 serial ports and a TCP/IP port, ? > attach 7 of the serial ports to each of the ES40s, then use a B > cross-over cable to attach the TCP/IP port to a cheap PC running? > Win98, start 7 telnet sessions on the PC, and connect each of 9 > them to one of the serial ports on the terminal server.  > > > I assume that I can find a terminal server that will do this > "reverse" telnet.  >  > Is anyone doing this?      Yes.  $ >What terminal server are you using?  
 decserver 700    > Any issues I should consider?   O No point in bringing a damn billybox or even tcp into the equation.  You can do & everything from the alphas with LAT.    O I basically connect each of the consoles to a port on the decserver, except for H two machines that are the core of my cluster.  These still have physicalM terminals.  To connect to any of the consoles for non-core machines, I simply O do "set host/dte ltaxxx", where ltaxxx is the port for a particular machine.  I M use a logical nodename_console rather than keeping up with port numbers.  The N following command procedure sets everything up (using 3 different decservers):  
 $ set noon $ lcp := $latcp & $ lcp create port lta602: /application& $ lcp create port lta603: /application& $ lcp create port lta604: /application& $ lcp create port lta605: /application& $ lcp create port lta606: /application& $ lcp create port lta607: /application& $ lcp create port lta608: /application $!& $ lcp create port lta702: /application& $ lcp create port lta703: /application& $ lcp create port lta704: /application& $ lcp create port lta705: /application& $ lcp create port lta706: /application $! $!& $ lcp create port lta802: /application& $ lcp create port lta803: /application& $ lcp create port lta804: /application& $ lcp create port lta805: /application& $ lcp create port lta806: /application $!8 $ lcp set port lta602: /node=ds700 /port=port_2/noqueued8 $ lcp set port lta603: /node=ds700 /port=port_3/noqueued8 $ lcp set port lta604: /node=ds700 /port=port_4/noqueued8 $ lcp set port lta605: /node=ds700 /port=port_5/noqueued8 $ lcp set port lta606: /node=ds700 /port=port_6/noqueued8 $ lcp set port lta607: /node=ds700 /port=port_7/noqueued8 $ lcp set port lta608: /node=ds700 /port=port_8/noqueued $! $!9 $ lcp set port lta702: /node=ds700b /port=port_2/noqueued 9 $ lcp set port lta703: /node=ds700b /port=port_3/noqueued 9 $ lcp set port lta704: /node=ds700b /port=port_4/noqueued 9 $ lcp set port lta705: /node=ds700b /port=port_5/noqueued 9 $ lcp set port lta706: /node=ds700b /port=port_6/noqueued  $! $! $!9 $ lcp set port lta802: /node=ds700c /port=port_2/noqueued 9 $ lcp set port lta803: /node=ds700c /port=port_3/noqueued 9 $ lcp set port lta804: /node=ds700c /port=port_4/noqueued 9 $ lcp set port lta805: /node=ds700c /port=port_5/noqueued 9 $ lcp set port lta806: /node=ds700c /port=port_6/noqueued  $! $!$ $ define/system harpo_console lta806$ $ define/system chico_console lta606& $ define/system groucho_console lta702$ $ define/system curly_console lta704    E I normally access the consoles one at a time.  I have a window called L "consoles" and connect as needed.  When I want to look at another console, IN disconnect with control backslash and set host/dte to a different lta device. G Since my machines rarely crash, this is sufficient.  I connect only for @ deliberate reboots.   If you want to look at all of the consolesD simultaneously, there is no reason you can't have a window for each.      A > I'm trying to do this without buying any special software (like  > Polycenter Console Manager).   None needed.   --  O =============================================================================== K Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachyon.xxx 8 http://www.tachyon.xxx/www/tachyon.html and wayne.html  K change .xxx to .com in addresses above, assuming you are not a spambot  :-) O =============================================================================== N Butler:"Gentlemen!"  Curly(as he and other Stooges look around):"Who came in?"   ------------------------------   Date: 24 Apr 2000 18:39:56 CST; From: wayne@tachyon.xxx.190136.killspam.03a7 (Wayne Sewell) < Subject: Re: "Reverse" Telnet as AlphaServer serial console?. Message-ID: <LadBV1k0BISY@tachxxsoftxxconsult>  c In article <012e01bfae06$3c43db20$020a0a0a@xile.realm>, "John E. Malmberg" <wb8tyw@qsl.net> writes: . > Dale A. Dellutri <ddellutr@enteract.company> > A >> I will soon have 7 separate ES40s, each of which need a system A >> console.  When I just 3, I used a VT520 with three sessions to A >> have three serial consoles with one screen.  Now, with 7, that  >> won't be practical. >>A >> I'm considering doing some sort of "reverse" telnet setup: use B >> a simple terminal server with 8 serial ports and a TCP/IP port,@ >> attach 7 of the serial ports to each of the ES40s, then use aC >> cross-over cable to attach the TCP/IP port to a cheap PC running @ >> Win98, start 7 telnet sessions on the PC, and connect each of: >> them to one of the serial ports on the terminal server. >>= >> Is anyone doing this?  What terminal server are you using?   >> Any issues I should consider? > M > I prefer to use separate terminals.  This could be a space problem for you, 1 > if you do not have room for another two VT520s.  > J > The only time that you need access to the consoles is when you are doingF > some software upgrades, standalone backup, and hardware diagnostics. > L > For all other operations, I use a VMS workstation with DECWindows-Motif toN > access the systems.  The instructions for setting up DECNET objects to allow$ > easy launching of remote DECterms. >  > G > There are two big drawbacks for using a terminal server on a Console.  > I > 1. If the Terminal Server is down, you  have no access to your systems.  > H > Your systems may not even boot if the terminal server is down.  If theL > terminal server does not have it's own flash ROM, it requires at least oneL > load host to boot it.  Just think, a full computer room power failure, andJ > you have no way of recovering.  Have you heard all the stories of people2 > accidentally hitting the emergency power button?  L This is why I use a hybrid setup, core machines with physical terminals, allK other machines through the decserver.  One of the core machines is the load M host for the decserver.  In the full power outage you describe, this is not a M problem.  You bring up the core machines first, the decserver loads, then you  bring up the others.     --  O =============================================================================== K Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachyon.xxx 8 http://www.tachyon.xxx/www/tachyon.html and wayne.html  K change .xxx to .com in addresses above, assuming you are not a spambot  :-) O =============================================================================== N Butler:"Gentlemen!"  Curly(as he and other Stooges look around):"Who came in?"   ------------------------------  % Date: Mon, 24 Apr 2000 20:42:40 -0400 2 From: "Richard B. Gilbert" <DRAGON@compuserve.com>* Subject: Re: Another VAX to Alpha question7 Message-ID: <200004242042_MC2-A26C-343D@compuserve.com>   J         The compiler will try to align things for you.  I think you would=  E have to force incorrect boundary alignment.  If you do something like C declare a struct with quadword, char, quadword, and don't allow the E compiler to insert seven bytes of filler, you are going to see SEVERE J performance problems every time you touch whichever of those quadwords th= e 6 compiler was unable to align, probably the second one.  F         You shouldn't have many instances of this sort of thing.  GoodF programmers have known, for the past thirty years or so, that boundaryF alignment makes things run faster.  Older CISC processors did boundaryE alignment with some grace and efficiency but the cost in on-chip real J estate is just too great in most RISC architectures.  I believe that just=  G about the entire fixup is done in software and it just isn't very fast.   $ Message text written by "John Nixon"H >I appreciate the comments I have received on VAX to Alpha conversion. I haveF been asked another question about this effort (and I am sure I will be asked  a lot more).  H One of the major warnings of areas to be concerned about have to do withG boundary allingments.  The question is whether these problems will show G themself during Compile or Link, or are we likely to see those problems  whenF the program is run.  Many of the problems we had with the VAXC to DECCJ Conversion did not show up until after the executables were built and run=  4 and then failed (frequently with access violations).  J I look forward to one day  riding the wave, rather than constantly paddli= ng
 to catch up.<    ------------------------------   Date: 25 Apr 2000 00:25:03 GMT From: ytsyjm@mindspring.com " Subject: Awesome Opportunity!  264- Message-ID: <8e2okv$i01223@magpie.cat.net.th>   + http://www.cashactive.com/go/?refid=AAA-893    ------------------------------  % Date: Mon, 24 Apr 2000 22:34:45 -0400 . From: Michael Austin <maustin@nc.prestige.net>9 Subject: Re: Catastrophic EVE/TPU documentation error :-) / Message-ID: <39050445.F61583FE@nc.prestige.net>   , This is a multi-part message in MIME format.& --------------08834F70C9BA99506BEE0622* Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit    Bob Koehler wrote: <snip>  8 > the other hande TPU didn't ship with ULTRIX very long.  f What a shame.. I would love to have TPU running on the various u**xes (Solaris, HPUX, AIX,Linux) etc..   Michael Austin Senior Rdb DBA Consultant    > H > ----------------------------------------------------------------------A > Bob Koehler                     | Computer Sciences Corporation ? > Hubble Space Telescope Payload  | Federal Sector, Civil Group G >  Flight Software Team           | please remove ".aspm" when replying   & --------------08834F70C9BA99506BEE0622- Content-Type: text/x-vcard; charset=us-ascii;   name="maustin.vcf"  Content-Transfer-Encoding: 7bit , Content-Description: Card for Michael Austin  Content-Disposition: attachment;  filename="maustin.vcf"    begin:vcard  n:Austin;Michael   tel;work:704-947-1089  x-mozilla-html:FALSE org:Michael Austin, Inc 
 adr:;;;;;; version:2.1 + email;internet:michaelaustininc@hotmail.com  title:President  x-mozilla-cpt:;0 fn:Michael E. Austin	 end:vcard   ( --------------08834F70C9BA99506BEE0622--   ------------------------------  % Date: Mon, 24 Apr 2000 16:08:13 -0500 , From: "Glenn C. Everhart" <Everhart@gce.com>+ Subject: Fall 1999 VMS/LT Sigtapes released ' Message-ID: <3904716D.21F8DE3C@gce.com>   8 The initial copies of the Fall 1999 VMS/LT sigtapes have7 been mailed to the tree head and other net distribution 2 points. Dar Schumann as usual is tree coordinator.  	 Contents:  Fall 1999 VMS/L&T SIG Tapes  ---- ---- ------- --- -----   H This material is all freely distributable, passed around without charge.  ? The following are contents of the respective directories in the  [VMSLT99B...] tree.   A [APACHE]                Apache web server for VMS with patches to *                         improve operation.D [ARPWATCH]              Watch ARP (Address Request Protocol) packets!                         from VMS. 5 [DJS]                   Info on burning CDs under VMS ' [ED]                    EDT like editor A [EWS]                   VAXEln windowing software for VAXstations H [GCE]                   Frag Avoider. Eliminates most disk fragmentationE                         before it happens; makes file allocation CBT,  alsoG                         files grow by a fraction of their current size.  Now =                         installs right on heavily used disks. F                         Also canonical VDdriver distributions for axp, vax.D [GNU]                   FSF software since 99A: new GCC, G77, Emacs, more. B [JED]                   New SLANG language interpreter for editorsF [MOREAU]                DECwindows utilities including Ghostscript (PSD                         interpreter), Mosaic (web browser), Timidity music 8                         editor, sound file to MIDI, etc.D [NET]                   Numerous tools. Includes Python for VMS, M4, Samba F                         use hints, XV, VMS backup reader, imagemagick,H                         Omega (Mumps clone), how to set up a certificateH                         authority in VMS, Vi clone, XTerminal update for VMS,"                         much more.C [NETHACK]               Game from the internet, related to Dungeon. F [NT]                    Many internals hints and tools for Windows NT.E [PERL]                  Latest Perl (string oriented utility language  and /                         system), builds on VMS. G [PHRACK]                Internals articles, mostly unix related, on how  to2                         attack and defend systems.: [PINE]                  Mail agent, MIME aware. VMS usable@ [SAMBA]                 System to let VMS or unix share files or printers4                         with Windows. Bidirectional.G [SDL]                   Structure Definition Language for VMS. Updated. D [SEC]                   Security relevant information and utilities. Much is .                         generic; many sources.F [SILU]                  Package of VMS utilities that is somewhat unix like. B                         Includes a VMS SED, Unarj, a BBS, a record extract,6                         touch, various others for VMS.F [.TK]                   CVTLIS - convert .LIS to source. JUMP - become another B                         user. MGFTP - VMS structure aware fast FTPH                         client/server. MMK - make or MMS clone. NBS- set yourH                         time to match NBS clock. REMIND - reminder util.F                         WAON - Convert WAV to MIDI file. (LONG step to?                         making a score from a sound recording.) = [TMESIS]                VMS internals advice from the master. D [VU]                    VMS utilities, many types. VMS BASH (shell), utils,H                         Ghostscript, Ghostview, diff/patch, xpdf, tecoc, more. 3 [WWW]                   Lynx text only web browser.    Thanks to contributors.  Glenn Everhart  + (sorry about the misdirected mail earlier.)    ------------------------------  % Date: Mon, 24 Apr 2000 16:36:41 -0400	+ From: Tim Shoppa <shoppa@trailing-edge.com>   Subject: ftp.decus.org progress?1 Message-ID: <39047819.11BBF137@trailing-edge.com>   6 Is there any news to report on the progress of getting= ftp.decus.org restored to operation?  It's been several weekse now...   Tim.   ------------------------------  # Date: Mon, 24 Apr 2000 20:56:58 GMT 0 From: "Terry C. Shannon" <shannon@world.std.com>$ Subject: Re: ftp.decus.org progress?& Message-ID: <FtJGDE.8wp@world.std.com>  8 "Tim Shoppa" <shoppa@trailing-edge.com> wrote in message+ news:39047819.11BBF137@trailing-edge.com... 8 > Is there any news to report on the progress of getting? > ftp.decus.org restored to operation?  It's been several weeks  > now...  8 This US Board of Directors member has no news to report.   regards,  
 terry shannon    ------------------------------  % Date: Mon, 24 Apr 2000 17:08:42 -0400 + From: Tim Shoppa <shoppa@trailing-edge.com>y$ Subject: Re: ftp.decus.org progress?1 Message-ID: <39047F9A.580EA6A4@trailing-edge.com>e   Terry C. Shannon wrote:s > : > "Tim Shoppa" <shoppa@trailing-edge.com> wrote in message- > news:39047819.11BBF137@trailing-edge.com...-: > > Is there any news to report on the progress of gettingA > > ftp.decus.org restored to operation?  It's been several weeksk
 > > now... > : > This US Board of Directors member has no news to report.  B Am I the only one to care?  I poked around DECUSserve a little bitA and didn't see anyone complaining there... maybe I'm the only oner who uses the library anymore?l   Tim.   ------------------------------  # Date: Mon, 24 Apr 2000 21:38:17 GMT  From: coy@eisner.decus.org$ Subject: Re: ftp.decus.org progress?' Message-ID: <2000Apr24.173817.1@eisner>   ! You are not the only one to care.s  H I do know that the situation is being studied.  Beyond that, I know very little.t  _ In article <39047F9A.580EA6A4@trailing-edge.com>, Tim Shoppa <shoppa@trailing-edge.com> writes:  > Terry C. Shannon wrote:t >> t; >> "Tim Shoppa" <shoppa@trailing-edge.com> wrote in message . >> news:39047819.11BBF137@trailing-edge.com...; >> > Is there any news to report on the progress of getting B >> > ftp.decus.org restored to operation?  It's been several weeks >> > now...  >>  ; >> This US Board of Directors member has no news to report.  > D > Am I the only one to care?  I poked around DECUSserve a little bitC > and didn't see anyone complaining there... maybe I'm the only onen > who uses the library anymore?w >  > Tim.   ------------------------------  % Date: Mon, 24 Apr 2000 14:46:15 -0400p. From: Michael Austin <maustin@nc.prestige.net>* Subject: Re: Help needed by non-VMS person/ Message-ID: <39049677.E7999913@nc.prestige.net>m  , This is a multi-part message in MIME format.& --------------48139FD69F6A4E1F175A2E38* Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit    Grant,  T Who is your Process Control Vendor?  I have seen applications written on VMS systemsU that gave snapshots of the real-time processing ever 1-3 minutes.  Some snapshot data V was loaded into a database every 10 minutes.  Using DECNET.  There are a lot of peicesX missing to your puzzle, so if you can gice us a few more details, we can give you better answers.   Michael Austin DBA Consutlant   kbrown780@isd.net wrote:  V > In article <sbvab6jree683@news.supernews.com>, "grant" <grantf@mistral.co.uk> wrote:* > > I just need a bit of guidance actualy. > >cO > > I work for an energy company in the downstream (non-drilling refining) sidenN > > of the business. We have a requirement to connect into the process controlP > > system of one of our electricity generating plants in order to see real-time( > > information on the generator output. > >rO > > The problem (for us) is that the control system in running VMS and the data C > > streams directly to the clinet nodes using the decnet protocol.: > >eP > > We are some 300 miles away so need to connect a system up via a leased line.M > > The main crux of my question is : Is the a PC client for VMS (pathworks?)lM > > that can enable me to receive real-time info from a VMS machine in decnetdN > > format, and additionally what is necessary from the comms side to get this > > link working.  > >l > > Cheers in advancer > >t	 > > Grant  > >c > >  >t > Hi Grant,e >76 > Can I ask who your process control system vendor is? >7 > --
 > Keith Browne > kbrown780@isd.netc  & --------------48139FD69F6A4E1F175A2E38- Content-Type: text/x-vcard; charset=us-ascii;0  name="maustin.vcf"e Content-Transfer-Encoding: 7bitd, Content-Description: Card for Michael Austin  Content-Disposition: attachment;  filename="maustin.vcf"l   begin:vcard  n:Austin;Michael q tel;work:704-947-1089e x-mozilla-html:FALSE org:Michael Austin, Inch
 adr:;;;;;; version:2.1i+ email;internet:michaelaustininc@hotmail.coms title:President  x-mozilla-cpt:;0 fn:Michael E. Austin	 end:vcard   ( --------------48139FD69F6A4E1F175A2E38--   ------------------------------  + Date: Mon, 24 Apr 2000 14:38:26 -0300 (EST)  From: becherini@vortex.ufrgs.bre; Subject: re: how do i kwno how much memory is on my system? , Message-ID: <00042414382662@vortex.ufrgs.br>  : Received:	by vortex.ufrgs.br (V5.0A-1, OpenVMS V7.2 Alpha)+ From:		Fabio Becherini <becherini@ufrgs.br>h 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 /UFRGS=O _______________________________________________________________________________n  ! . From: joecarlos@yahoo.com (Joe) 9 . Subject: how do i kwno how much memory is on my system?h% . Date: Mon, 24 Apr 2000 17:01:09 GMTs . To: Info-VAX@Mvb.Saic.Com  . D . is there a commnad that tells me the physical memory of my system?     	yes, "show memory" ...e  	 	regards,   N  _____________________________________________________________________________O |                                                                             |(O | Fabio Becherini                   System & Network Manager, Webmaster UFRGS |sO | CPD-UFRGS                         Centro de Processamento de Dados da UFRGS |2O |                                   Universidade Federal do Rio Grande do Sul |tO |                                                   Divisao de Rede e Suporte | O |                                          (55)(51) 316-5041 / 331-1215 (fax) |oO | Rua Ramiro Barcelos, 2574  -  Santa Cecilia  -  Porto Alegre - RS -  Brasil |nO |_____________________________________________________________________________|pO |                                                                             |tO | Cia-INFO (c) Ophicin@ das Informacoes                   Coordenacao Central |oO |_____________________________________________________________________________| O |                                                                             |lO | INTERnet:  fabio.becherini@ufrgs.br              DECnet:  vortex::becherini |AO |_____________________________________________________________________________|l   ------------------------------   Date: 24 Apr 2000 18:12:46 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman); Subject: Re: how do i kwno how much memory is on my system?p6 Message-ID: <8e22qu$1f5$1@mailint03.im.hou.compaq.com>  W In article <8F20856F0joeyahoocom@news.supernews.com>, joecarlos@yahoo.com (Joe) writes:eC :is there a commnad that tells me the physical memory of my system?   H   The SHOW MEMORY command will show you how much of the physical memory I   is configured by OpenVMS.  This value is equal or less than the amount sF   of physical memory present -- in most (but not all) cases, the valueJ   returned by SHOW MEMORY will be the amount actually configured.  (PleaseL   see the OpenVMS FAQ for related details in section MGMT11.)  To determine I   the amount of physical memory actually present in the system, you will rI   want to use one of various console-specific commands.  As you have not gL   told us which OpenVMS version and which OpenVMS platform -- please always L   remember to include this information whenever posting a question -- it is L   difficult to provide you with the specific console command(s) involved in J   probing the current syste, physical memory configuration.  Usually, thisG   involves a console SHOW command of some sort, but the console command=9   syntax and the console command set can and does vary...=  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Tue, 25 Apr 2000 02:06:53 GMT=$ From: Ed Wilts <ewilts@mediaone.net>; Subject: Re: how do i kwno how much memory is on my system? , Message-ID: <3904FDBD.4ED0B33C@mediaone.net>   Hoff Hoffman wrote:- > Y > In article <8F20856F0joeyahoocom@news.supernews.com>, joecarlos@yahoo.com (Joe) writes:RE > :is there a commnad that tells me the physical memory of my system?g > I >   The SHOW MEMORY command will show you how much of the physical memory J >   is configured by OpenVMS.  This value is equal or less than the amountH >   of physical memory present -- in most (but not all) cases, the valueD >   returned by SHOW MEMORY will be the amount actually configured.   F I'll definitely agree with this statement.  I "upgraded" a VS4000-90 aG year or two ago by simply autogen'ing the system and having it actuallyoH configure all of the physical memory present.  It turns out that someoneE had incorrectly set physicalpages to a number lower than what the boxtD had (probably a leftover from an earlier hardware configuration).  IE gave my customer an additional 8MB with a simple reboot.  When you'rei@ talking 24 to 32, that's well over 30% additional usable memory!   	.../Edh   -- e Ed Wilts Mounds View, MN, USA mailto:ewilts@mediaone.net   ------------------------------  % Date: Mon, 24 Apr 2000 14:53:50 -0400i1 From: Marc Lippmann <marc@lippmann.spamsucks.com> / Subject: Re: LK46W-A2 & "DO" key in Reflectionsh8 Message-ID: <d069gsokjr5igldf4s1beiubl2831v1cmc@4ax.com>  D Darren, are you having this problem all within one session, or is it9 just the ability to save the settings that's not working?    ------------------------------  % Date: Mon, 24 Apr 2000 17:03:01 -0500r* From: WILLIAM WEBB <WWEBB1@email.usps.gov> Subject: No Subjectn- Message-ID: <0033000021378034000002L042*@MHS>    =0Ai        UNSUBSCRIBE INFO-VAXrH ---------------------------- Forwarded with Changes -------------------= --------  From: WILLIAM W WEBB at RANC023L Date: 4/24/00 9:58AM% To: Info-VAX@Mvb.Saic.Com at INTERNETc5 Subject: Re[2]: Calculating Discordian dates from DCL5H -----------------------------------------------------------------------=	 --------=2   ------------------------------   Date: 24 Apr 2000 21:18:23 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: No Subject,6 Message-ID: <8e2dmv$5c2$1@mailint03.im.hou.compaq.com>  Z In article <0033000021378034000002L042*@MHS>, WILLIAM WEBB <WWEBB1@email.usps.gov> writes:  C   INFO-VAX subscription requests should be sent to the mail server, C   not to the mailing list.  Please see the OpenVMS FAQ for details,eE   or (if you still have it) the response that you received back when h%   you first subscribed to INFO-VAX...   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 24 Apr 2000 15:09:59 -0400-# From: Jim Agnew <agnew@hsc.vcu.edu>  Subject: Re: NT Breaks Again+ Message-ID: <39049C07.293DBFE4@hsc.vcu.edu>S  D yup... the guy is right!!!  (no wonder Carl Lydick's alter ego was a
 Kzinti...)   j   " Shane.F.Smith@healthnet.com wrote: > R > It was only the first edition of Ringworld where the ring rotated the wrong way.L > The second editon and onwards have that and a few other technical glitchesP > fixed. It's not until "The Ringworld Engineers" that he gets the attitude jetsR > in place, that one was found by a bunch of physics students who were working out > the numbers as an exercise.d > A > Shane (who actually reads the author's intro section sometimes)i > 9 > Jim Agnew <agnew@hsc.vcu.edu> on 04/24/2000 07:44:01 AMr >  >  >  >  >  >  > ! >  To:      Info-VAX@Mvb.Saic.Com- > * >  cc:      (bcc: Shane F Smith/WH/HN/FHS) >  >  >  >  Subject: Re: NT Breaks Againr >  > @ > hahahahahaahaha.......   shades of Ringworld, by Larry Niven.. >  > ever read it?? >  > j  > * > Andrew Harrison SUNUK Consultancy wrote: > >  > > Art Rice wrote:e > >d% > > > Ya just gotta love NT security:a > > >a7 > > > http://news.cnet.com/news/0-1003-200-1707928.html  > > >  > > > -- > > > Art Rice   *#-) > > > Special Data Processing Corporationt, > > > --------------------------------------, > > > All opinions expressed are mine and do+ > > > not reflect the views of my employer.A > > B > > On a simular vein this is a bug report from Microsoft that may > > amuse you. > > L > > check out http://support.microsoft.com/support/kb/articles/Q131/1/09.asp > > 1 > > It will the the Microsoft Flat Earth next :-)  > >  > > Regardsn > > Andrew Harrison] > > Enterprise IT Architecte   ------------------------------  % Date: Mon, 24 Apr 2000 16:48:40 -0400 0 From: JF Mezei <jfmezei.spamnot@vl.videotron.ca> Subject: Re: OpenVMS marketing/ Message-ID: <3904B320.790A245E@vl.videotron.ca>o   Rob Young wrote:J >         Maybe you would have more impact if you got your facts straight.G >         Sustaining membership costs $40 for a year.  Basic membership  >         is FREE.  J Since there is no longer a "DECUS" in Canada, since it is unclear what theN unlected, self-appointed and inactive board of the organisation that was meantL to replace it are doing, and since DECUS-USA does not (do they now ?) acceptL members from other countries, it is hard to know what will happen on June-18 when my PAKs will expire.   L Can I use my DECUS-Canada member number to renew even though DECUS Canada no longer exists ?t   ------------------------------  # Date: Mon, 24 Apr 2000 23:11:21 GMTi) From: "Neil Rieck" <n.rieck@sympatico.ca>  Subject: Re: OpenVMS marketing; Message-ID: <tw4N4.71410$2D6.1854411@news20.bellglobal.com>   ; JF Mezei <jfmezei.spamnot@vl.videotron.ca> wrote in message ) news:3904B320.790A245E@vl.videotron.ca...  > Rob Young wrote:L > >         Maybe you would have more impact if you got your facts straight.I > >         Sustaining membership costs $40 for a year.  Basic membershipr > >         is FREE. > L > Since there is no longer a "DECUS" in Canada, since it is unclear what theJ > unlected, self-appointed and inactive board of the organisation that was meantaG > to replace it are doing, and since DECUS-USA does not (do they now ?)M acceptF > members from other countries, it is hard to know what will happen on June-18  > when my PAKs will expire.o > K > Can I use my DECUS-Canada member number to renew even though DECUS Canada. no > longer exists ?    FYI all concerned,  H When you surf to http://ww.decus.ca you are directed to something calledH "The Canadian Association of Compaq Users". There is no mention of a twoI tier membership scheme, only C$5.00 for students and C$50.00 for everyonea else.   J Now with my luck, I'll fork over 50 bucks to get a CACU membership only toB discover that the OpenVMS hobbyist folk have never heard about theJ organization :) With similar potential foo-bars in Germany, France and anyL other members of the United States of Europe, I can only assume that someoneG is making it difficult for hobbyists to get a free license (yeah, yeah;i' always with the conspiracy theories...)t  H Seriously, it only took 2 minutes and a credit card to order a Solaris-8D CD-ROM from the Sun web site (if I didn't want a CD-ROM I could haveL downloaded Solaris-8 for free). I'm SURE that the folks at Compaq could keepI their costs down by also running an automated site (or at least selling ahJ US$99 version of OpenVMS from their corporate site like they're doing with Tru64)  I p.s. for any of you not familiar with the term "United States of Europe",lL many Canadians use it to describe the 390+ million consumers who make up EEC" (the large market across the pond)  
 Neil Rieck* Kitchener(New Berlin?)/Waterloo/Cambridge, Ontario, Canada.! http://www3.sympatico.ca/n.rieck/e6 http://www3.sympatico.ca/n.rieck/docs/alpha_diary.html   ------------------------------  # Date: Mon, 24 Apr 2000 23:45:57 GMT ) From: "Neil Rieck" <n.rieck@sympatico.ca>, Subject: Re: OpenVMS marketing< Message-ID: <V05N4.144293$1C2.3497135@news20.bellglobal.com>  9 Terry C. Shannon <shannon@world.std.com> wrote in message   news:FtIzKG.Ey2@world.std.com... >c [snip] >nG > Still, it's significant to note that the OpenVMS Hobbyist License waswH > available long before the Tru64 Technology Enthusiast License. And theK > Dallas-Fort Worth Compaq Users Group is largely responsible for the birth  of > the OpenVNS Hobbyist License.  >r   To all concerned,-  E Kudos to all those responsible for  the birth of the OpenVMS HobbyistiK license. But the world is changing very quickly and stuff that used to costeL big bucks is now either cheap or free. Solaris (a very respectable UNIX fromF Sun) used to be fairly expensive but now SPARC and Intel (for your PC)J versions can be downloaded from the Sun web site for free (or US$75 if youL want a CD-ROM). Many people think BeOS is the answer to every problem but ifH you want to check out the product yourself, you can download it for free from the Be Inc. web site.  K On the flip side, what is the meaning of free? LINUX is supposedly free and K yet the most popular way to get a LINUX CD-ROM is to buy a book with one in B the back which will set you back between C$25 and C$50. Windows isH supposedly NOT free and yet the last two PCs I purchased included a freeI legal Windows-9x CD-ROM (OK. So this is a bad example but I think you cant
 see my point)o   ...philosophical pause...e  J In the computer world today free means anything between $0 and $99 and the1 Tru64 introduction package meets the requirement.-   ...philosophical pause....  I Rhetorical question: what ever happened to "demand and supply" curves? OheK yeah that's right; Digital ignored them and that's why they almost went outsI of business. If the supply curves says you have to drop the price to $99,a then you've got to do it.e  
 Neil Rieck* Kitchener(New Berlin?)/Waterloo/Cambridge, Ontario, Canada.! http://www3.sympatico.ca/n.rieck/o6 http://www3.sympatico.ca/n.rieck/docs/alpha_diary.html   ------------------------------  % Date: Mon, 24 Apr 2000 20:42:38 -0400e2 From: "Richard B. Gilbert" <DRAGON@compuserve.com>8 Subject: Problem with upgrading an Alpha Server to 7.2-17 Message-ID: <200004242042_MC2-A26C-343C@compuserve.com>:  7         You didn't mention running AUTOGEN!  Did you? =9  J $ @SYS$UPDATE:AUTOGEN SAVPARAM REBOOT   !Not best but simplest to explain= .-    $ Message text written by Shawn MaloneG >I began upgrading a DEC 3000 Alpha Server on Thursday from OpenVms 7.1 A to 7.2-1.  The server was running UCX 4.2 and DecNet Phase IV.  =t    F I ran the upgrade and the OS portion loaded without a problem.  I then; upgraded UCX to TCPIP050A, again no problem, I have TCP/IP.e  : I applied the VMS721_UPDATE-V0100 patch, again no problem.  A I then set out to upgrade my DecNet to DECNET-PLUS-V7_2.  This isgB where the problem began.  The first thing it did was complain that= GBLSECTIONS, GBLPAGES, GBLPAGFIL and KSTACKPAGES were to low.s  B I was able to modify the first 3, but it is not letting me set theB KSTACKPAGES.  I have tried editing the ModParams.dat file and haveG included both MIN_KSTACKPAGES=3D2 and KSTACKPAGES=3D2.  Then reboot theqC system.  When I look at tha params after the reboot it goes back tor one.  ? DecNet is giving me the error Net$Startup-F-NOREQSYSGEN, SYSGEN " parameters not at required values.  : I have also installed the DNVOSI072ECO01 patch for Decnet.  > When installing this file the firs thing that happened was the following error:  7 The following product will be installed to destination: F     DEC AXPVMS DNVOSIECO01 V7.2            DISK$ALPHASYS:[VMS$COMMON.]B %PCSI-I-RETAIN, file [SYSLIB]DTSS$SHR.EXE was not replaced because) file from kit has lower generation numberc  C Can someone please give me some ideas of what I need to do next ande0 what I need to do to get the KSTACKPAGE to hold. <    ------------------------------  # Date: Tue, 25 Apr 2000 01:38:09 GMT 0 From: "Terry C. Shannon" <shannon@world.std.com> Subject: Recent SKC Postings& Message-ID: <FtJtEn.L8J@world.std.com>  8 Be advised that new SKC postings are now available at...     http://www.acersoft.comn   enjoy,   terry s7   ------------------------------   Date: 24 Apr 2000 14:54:27 PDTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515), Subject: Re: system disk move and batch jobs3 Message-ID: <nCdLEJjt6ra3@mccdev.slac.stanford.edu>n  H In article <eWYL4.27293$WF.1111865@bgtnsc04-news.ops.worldnet.att.net>, 5     "John Nixon" <jorlnixon@worldnet.att.net> writes:6N > We will be moving our system disk from a host based shadow set of  a pair ofM > third party disks to a StorageWorks hardware based stripe/shadow set (usingwL > an image copy).  We are going to retain the same shadow device designationN > (DSA9)  on the new system disk so that we may shadow this to another similarL > hardware stripe/shadow set on a different pair of Storageworks controllers > later.  H         OK, now I'm a little confused...  On the one hand you say you'reH     moving _from_ an  HBVS  _DSA_  device  _to_  a controller (hardware)H     based  mirror/stripe  set (that's _mirroring_ when  it's  controllerH     based) which gives you a (single) _DUA_  (or  _DKA_?)  device.   ButH     then  you  say you want to retain the _DSA_ "designation".  StrictlyH     speaking, that's out of  your  control.   Of course, there's nothingH     stopping  you  from  defining  "DSA9" to point to  the  new  device,H     /system/exec,  in  your  system  startup.   That  should   make   it/     transparent to the rest of your procedures.   H         Also note that it is  preferable  to  stripe mirror sets, ratherH     than  mirroring  stripe  sets.  If you lose a disk  from  a  stripedH     mirror set, only that single disk needs  to  be  mirror-copied  backH     into  its mirror set while the stripe set marches merrily along.  IfH     you lose a disk from  a  mirrored  stripe  set, the whole stripe setH     drops  out  of the mirror set and must be rebuilt with the new  diskH     before it, the whole stripe set, can be mirror-copied back into  the     mirror set._  H         FWIW, I also would  have  a  small  question  about striping theH     system  disk.   Do you really need extra space there (which you  getH     with striping) given large size of currently  available  disks?   IsH     there some performance issue that you think striping will help with?$     Just something to think about...  J > Will batch jobs that are holding, and which reference input files on theI > original DSA9 run properly on the new DSA9, or will they all have to be_K > resubmitted.    We have a lot of them and testing is not a possibility at- > this time.  I         It should be transparent given you define DSA9 in SYLOGICALS.COM.                -Ken c -- :M  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edul:  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: 24 Apr 2000 22:10:38 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman), Subject: Re: system disk move and batch jobs6 Message-ID: <8e2gou$6l8$1@mailint03.im.hou.compaq.com>   In article <nCdLEJjt6ra3@mccdev.slac.stanford.edu>, Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515) writes:I :In article <eWYL4.27293$WF.1111865@bgtnsc04-news.ops.worldnet.att.net>, a6 :    "John Nixon" <jorlnixon@worldnet.att.net> writes: : K :> Will batch jobs that are holding, and which reference input files on theoJ :> original DSA9 run properly on the new DSA9, or will they all have to beL :> resubmitted.    We have a lot of them and testing is not a possibility at
 :> this time.u :yJ :        It should be transparent given you define DSA9 in SYLOGICALS.COM.  I   Yes, in most cases a logical name (/EXEC/SYSTEM) will work just fine...-H   If you can/do move away from physical device names, I would recommend ;   moving to DISK$volumelabel or similar device reference...z  L   That said, there are things that are not quite as transparent (Oracle Rdb N   databases can be a salient example), and I sometimes end up configuring and N   running a one-volume host-based shadowset using a disk on a RAID controller O   for just this reason -- to have a device name (DSAmumble) that permits me to rM   easily reconfigure storage "underneath" the particular software involved...m  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Mon, 24 Apr 2000 23:00:23 GMT / From: "John Nixon" <jorlnixon@worldnet.att.net>l, Subject: Re: system disk move and batch jobsG Message-ID: <bm4N4.31527$fV.1905804@bgtnsc05-news.ops.worldnet.att.net>   L I am sorry I confused you.  The answer to my question was that an Image copyD retains FID, thus the batch jobs would work since I retained the DSAH numbering scheme.  I mistakenly thought a file would get a new FID on an image copy to a larger disk.  ' In response to your other observations:r: I do stripe mirror sets rather than mirroring stripe sets.  F I do get better performance by striping two 4.3GB mirror sets, and theJ resultant 8.5GB array is not too big.  We have several large dump files on@ the system disk and I like to keep a lot of free space to reduceG fragmentation.  It is the extra performance I get from striping 7200rpmAK disks that allows me to keep accounting files, authorization files etc.. on@G the system disk. (maybe a new disk spinning at 10k would do fine, but Ir dont' have that).i  E Even though the HSJ array presents a single DU device to VMS,  I have-J SHADOW_SYS_DISK and SHADOW_SYS_UNIT defined which puts the DSA designationJ back in my hands.  By doing this, I can later  use HBVS across controllersF (or data centers).  I also keep DSA device numbers under my control byI mounting the DU devices presented by the HSJ with a MOUNT/SHADOW command.     2 "Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515". <Fairfield@SLAC.Stanford.EDU> wrote in message- news:nCdLEJjt6ra3@mccdev.slac.stanford.edu...-I > In article <eWYL4.27293$WF.1111865@bgtnsc04-news.ops.worldnet.att.net>,-7 >     "John Nixon" <jorlnixon@worldnet.att.net> writes:3H > > We will be moving our system disk from a host based shadow set of  a pair of H > > third party disks to a StorageWorks hardware based stripe/shadow set (usingB > > an image copy).  We are going to retain the same shadow device designationsH > > (DSA9)  on the new system disk so that we may shadow this to another similar.B > > hardware stripe/shadow set on a different pair of Storageworks controllersn
 > > later. >eJ >         OK, now I'm a little confused...  On the one hand you say you'reJ >     moving _from_ an  HBVS  _DSA_  device  _to_  a controller (hardware)J >     based  mirror/stripe  set (that's _mirroring_ when  it's  controllerJ >     based) which gives you a (single) _DUA_  (or  _DKA_?)  device.   ButJ >     then  you  say you want to retain the _DSA_ "designation".  StrictlyJ >     speaking, that's out of  your  control.   Of course, there's nothingJ >     stopping  you  from  defining  "DSA9" to point to  the  new  device,J >     /system/exec,  in  your  system  startup.   That  should   make   it1 >     transparent to the rest of your procedures.e >aJ >         Also note that it is  preferable  to  stripe mirror sets, ratherJ >     than  mirroring  stripe  sets.  If you lose a disk  from  a  stripedJ >     mirror set, only that single disk needs  to  be  mirror-copied  backJ >     into  its mirror set while the stripe set marches merrily along.  IfJ >     you lose a disk from  a  mirrored  stripe  set, the whole stripe setJ >     drops  out  of the mirror set and must be rebuilt with the new  diskJ >     before it, the whole stripe set, can be mirror-copied back into  the >     mirror set.h >.J >         FWIW, I also would  have  a  small  question  about striping theJ >     system  disk.   Do you really need extra space there (which you  getJ >     with striping) given large size of currently  available  disks?   IsJ >     there some performance issue that you think striping will help with?& >     Just something to think about... >oL > > Will batch jobs that are holding, and which reference input files on theK > > original DSA9 run properly on the new DSA9, or will they all have to beaJ > > resubmitted.    We have a lot of them and testing is not a possibility at > > this time. >pK >         It should be transparent given you define DSA9 in SYLOGICALS.COM.  >  >             -Ken > --/ >  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:      650-926-3515L >  ------------------------------------------------------------------------- ----D >  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  # Date: Mon, 24 Apr 2000 20:31:51 GMTe/ From: "John Nixon" <jorlnixon@worldnet.att.net>c/ Subject: Re: The Movie "Breaking Point" + PDP11tG Message-ID: <Xa2N4.33023$WF.1516771@bgtnsc04-news.ops.worldnet.att.net>o  L Does anyone know the market share that Alpha has in Hollywood?  Unless it isI very high, it would be worse to publicize just a few because people wouldw$ notice the omission in the majority.  J I am still embarassed everytime I hear Rick Marcello brag about visibilityL by pointing to the Northern Lights advertisement.  I have seen scores of SunL ads and countless IBM and HP ads since then.  VMS marketing is still sitting on its LAUREL. (singular).    6 "David Turner" <d_b_turner@yahoo.com> wrote in message) news:sg7b57c5cdo172@corp.supernews.com...rL > Anyone ever noticed the title in the beginning of the movie Breaking Point ?  >tD > In the "music by" section it gives some guy's name with "his PDP11	 computer"  >  > Wow !u >i0 > Shame they never pushed Alpha's that way huh ? >i > How about this:F >.E > "Indepedence day  - rendering by Digital Alpha Technology  Personala > Workstation 433a+ > Call 1 800 Digital for sales information"o >nI > Instead it was Shhhhhhh... don't let them know they were using Linux one our G > Alpha's - in fact, don't let them know they were using Alpha's at alln >w > Can we say Pathetic ?????  >h > -- > David Turner! > Island Computers US Corporation  > 2700 Gregory Streete > Savannah GA 31404d > Tel: 912 447 6622t > Fax:912 201 0096 > sales@islandco.com >o >t   ------------------------------  % Date: Mon, 24 Apr 2000 13:59:29 -0400t0 From: Glenn and Mary Everhart <Everhart@gce.com>( Subject: Re: Using DiskMizer with Oracle' Message-ID: <39048B81.1CB88FF4@gce.com>   ? The major problem with using a compressing package like this toh= store databases that are getting updated is that because diskh< is compressed in chunks of, say, 16 blocks or 32 blocks, any@ error on disk that develops affects not one block, but the whole> 16 or 32. Since the compressing disk doesn't know what the appA is storing on disk, it can't protect you in any way from an index @ record, internal bitmap, or other structure from being corrupted along with a data block.  ; Consider: block 1000 of your disk gets corrupted; maybe theO; disk got too hot, or the SCSI bus glitched, or whatnot. The < database will generally have a transaction log that will let= it recover from the write error on 1000 (provided the packageO; returns prompt write errors to the app). However, the index6@ record in block 1008 hasn't been touched in months as far as the> database knows, but is now toast along with block 1000 and the8 other blocks in the area, because the data error gives a* chunk of disk that won't decompress right.  @ Now your index is corrupted. If your disk package returns errorsA when the underlying disk fails, you might find out that something-B went wrong. If not, not. However you won't know what the extent ofB possible damage to your database is without analyzing whatever was3 in the vicinity of a failure physically. Good luck.0  > With the cost of storage what it is I'd recommend just getting< bigger disks. I've seen 50GB scsi drives for $900 or so with case and power supply.  C For archival storage, cmphighc.zip on the sigtapes awhile back willfB let you mount a compressed disk image as a read only disk too, butA you should keep offline backups of such beasts. You cannot be too ? paranoid in how you trust computer storage, particularly onlineo* storage. If nothing else, heat can get it!  = The compressing virtual disk can compress anything, databasesu= included, and can do a very good job; the vulnerability noteda< is inherent in undrelying media not being eternal and in the? files for databases having structure of which the virtual disk o@ is unaware. If you plan to use such a solution, be sure you haveE regular backups, and keep them a long time since corruption such as Ie& mention can go undetected a long time.     Bill Todd wrote: > / > Ed Burke <clegg67@yahoo.com> wrote in messagen; > news:20000419190208.11661.qmail@web3003.mail.yahoo.com...d >  > ...  > 4 > > In theory, DiskMizer shouldn't get in the way of4 > > accessing the disk.  It does the compression and5 > > decompression on the fly, transparent to VMS.  In 7 > > fact, since it reduces I/O, it could even be fastera9 > > than a regular disk.  (In exchange for using some CPUn2 > > to handle the compression, you only write (for8 > > example) half as much to disk if you are getting 2:1: > > compression).  We have found this to be true with very > > large text files,p > G > Compressing a large sequentially-accessed file is an optimal case foraL > compression:  the compressed data remains contiguous and in the same orderN > on disk.  Compressing randomly-accessed data such as database pages is a lotI > more iffy:  that data *may* be accessed sequentially in table scans, so N > should ideally remain contiguous and in-order (and Oracle will likely try toN > keep it so), but if something else is revectoring the data based on where itA > will 'fit' after random-update compression this may not happen.a > J > There can certainly be exceptions, but my inclination is to believe thatL > compression is primarily useful for sequentially-accessed data that is not > modified in place.  YMMV.n >  > - bill > # >  and Symark says it would be trueP: > > with Oracle, but some confirmation (or rejection) from8 > > others who have gone this route could help us decide# > > whether to pursue this further.  > >m > >   Ed > >16 > > __________________________________________________ > > Do You Yahoo!?0 > > Send online invitations with Yahoo! Invites. > > http://invites.yahoo.com   ------------------------------  % Date: Tue, 25 Apr 2000 10:19:48 +12000* From: "Ryan C. Price" <pricerc@ihug.co.nz>3 Subject: Re: VAX CI storage vs served fibre channel * Message-ID: <8e2ha6$b94$1@news.ihug.co.nz>  B Even if FC only delivers 10% of its theoretical performance, it'll? still deliver more than your current 7MB/s. And I'm sure with ao< decent FC implementation, you'll exceed 10% quite easily. By? decent, I mean properly designed by a storage expert; there are ? many ways to build an FC network, and many ways to do it badly.a  @ Note that there are some significant issues with quality in some< FC implementations, and a lot of vendor FUD and over-stating? of capabilities. Being a former StorageWorks consultant, I feeltC I can trust their technology (they are mostly of the old DEC-schooln( of quality engineering being paramount).  @ As an example of a bad design; some vendors are passionate aboutB flogging FC disks, and will sell you an array of 20 FC disks on anA FC loop, which will give you pathetic performance, of course theyaD wouldn't tell you that, they'll just tell you how wonderful and fast FC is.   /Ryano3 (no I don't work for Compaq, at least not anymore).o    C "Martin Oettl" <Martin.Oettl@drop-this.compaq.com> wrote in message:. news:39001124.676A028C@drop-this.compaq.com... > John Nixon wrote:: >c	 > > Ryan,t > > E > > How do I get multi-site shadow sets without host based shadowing.n That7 > > would appeal to me, but I have never heard of that.o > > F > > As for the math,  that part is simple, if you believe that in real life the? > > math numbers hold up.  You only get the theoretical thruput  numbers if youC > > eliminate every bottleneck.  With CI I have come close and  cana actuallyE > > sustain 7MB/second, but I am not sure what kind of config I wouldo need toe! > > maintain FC thruput maximums.l > >@ >l > Data Replication Manager:w >  >mF http://www.compaq.com/products/storageworks/Storage-Management-Softwar e/DataRepindex.htmlo >e > Martin >e   ------------------------------  # Date: Tue, 25 Apr 2000 00:08:59 GMTi/ From: "John Nixon" <jorlnixon@worldnet.att.net>n3 Subject: Re: VAX CI storage vs served fibre channeloG Message-ID: <vm5N4.21628$PV.1525956@bgtnsc06-news.ops.worldnet.att.net>w  > Thanks for your comments.  Let me throw in a few more details., (most of which were in my original posting).  @ My question was not so much the performance of FC, which I agreeE is excellent, and further I have faith that the Compaq implementation  is probably among the best.(  B However, my concern is regarding serving disk storage to my VAXes.@ My VAXes currently connect to my disk storage via 4 CIXCDs each.5 Each CIXCD is attatched to one of four Star Couplers.   G I frequently see sustained agregrate thruput of greater than 20 MB/sec. - Sometimes as much as 25MB/second to my VAXes.h  H In this VMS Cluster, I also have an Alpha 8400 , and I may swing anotherD similar Alpha.  The 8400 has 4 CIPCAs that attatach to the same four star couplers as the VAXes.p  B Now, if I attach the Alpha to a FC Storage Area Network, and I useH the Alpha to server storage to the VAXes, what thruput will I be able toD maintain for the VAXes.  Would the storage be served via FDDI or CI?5 Would it divide the IO traffic equally between 4 CIs?   G So you see, my original question was not merely if FC is any good.  The 2 question was if it was viable in my configuration.    5 "Ryan C. Price" <pricerc@ihug.co.nz> wrote in messageo$ news:8e2ha6$b94$1@news.ihug.co.nz...D > Even if FC only delivers 10% of its theoretical performance, it'llA > still deliver more than your current 7MB/s. And I'm sure with ay> > decent FC implementation, you'll exceed 10% quite easily. ByA > decent, I mean properly designed by a storage expert; there areuA > many ways to build an FC network, and many ways to do it badly.u >iB > Note that there are some significant issues with quality in some> > FC implementations, and a lot of vendor FUD and over-statingA > of capabilities. Being a former StorageWorks consultant, I feel E > I can trust their technology (they are mostly of the old DEC-school * > of quality engineering being paramount). >|B > As an example of a bad design; some vendors are passionate aboutD > flogging FC disks, and will sell you an array of 20 FC disks on anC > FC loop, which will give you pathetic performance, of course they-F > wouldn't tell you that, they'll just tell you how wonderful and fast > FC is. >s > /Ryan 5 > (no I don't work for Compaq, at least not anymore).A >2 > E > "Martin Oettl" <Martin.Oettl@drop-this.compaq.com> wrote in messagec0 > news:39001124.676A028C@drop-this.compaq.com... > > John Nixon wrote:0 > >h > > > Ryan,> > > >aG > > > How do I get multi-site shadow sets without host based shadowing.D > That9 > > > would appeal to me, but I have never heard of that.r > > >eH > > > As for the math,  that part is simple, if you believe that in real
 > life theA > > > math numbers hold up.  You only get the theoretical thruputd > numbers if youE > > > eliminate every bottleneck.  With CI I have come close and  canA
 > actuallyG > > > sustain 7MB/second, but I am not sure what kind of config I wouldi	 > need tos# > > > maintain FC thruput maximums.  > > >h > >b > > Data Replication Manager:D > >n > >GH > http://www.compaq.com/products/storageworks/Storage-Management-Softwar > e/DataRepindex.htmlu > >d
 > > Martin > >p >i >    ------------------------------    Date: 24 Apr 2000 13:19:57 -0700* From: dunnett@mala.bc.ca (Malcolm Dunnett) Subject: RE: Verify of Backups, Message-ID: <lwo3jTt6eIST@malvm2.mala.bc.ca>  , In article <B529EC54-4A459@165.247.46.172>, 9    "Robert Deininger" <rdeininger@mindspring.com> writes:  > C >>One plausible theory to reconcile the two sets of results is thateF >>BACKUP implements the "skip to beginning of this save set" for largeB >>save sets as a rewind followed by a massive forward IO$_SKIPFILEB >>operation.  On my hardware, the device driver emulates this as aC >>series of skip record operations and counts one I/O per tape markcG >>found (three tape marks per save set is standard).  On your hardware, E >>the device driver ships the skip file request straight to the drives" >>which then accelerates the seek. > E > I'm getting more confused.  When I used MKSET to force fast skip ons> > and off, I wasn't able to see any difference in Backup.  ButD > SET MAGTAPE/SKIP=FILE was faster with fast skip.  I'll have to dig2 > up those results and see if they make any sense. > C    This was discussed on the list about a year ago. IIRC one of the B backup developers commented at the time that BACKUP never uses theC "skip file" operation ( due to it not working properly with certain.G devices? ), but instead requests a large number of record skips ( whichsD will abort when a file mark is seen ). Thus the MKSET option doesn't@ have any affect on BACKUP, though the impact on SET MAGT/SKIP is quite dramatic.p  C    To further add to the confusion it seems that some drives ( suchi@ as HP DATs ) will automagically translate a "skip huge number ofL records" request into a fast skip to the end of the file, but the DLTs won'tC (which was the cause of my earlier comment that /VERIFY of multipleW. savesets was much faster on DAT than on DLT ).  M ============================================================================= M Malcolm Dunnett      Malaspina University-College   Email: dunnett@mala.bc.caeH Information Systems  Nanaimo, B.C. CANADA V9R 5S5     Tel: (250)755-8738   ------------------------------   End of INFO-VAX 2000.230 ************************