Fixed-sized provisioning

Two traditional fixed-size host-based volume management methods typically are used on open-systems to organize storage space on a server. One method is the direct use of physical volumes as devices for use either as raw space or as a local or clustered file system. These are fixed-size volumes with a fixed number of disks, and as such, each has a certain inherent physical random input/output operation per second (IOPS) or sequential throughput (megabytes per second) capacity. A System Administrator manages the aggregate server workloads against them. As workloads exceed the volume's available space or its IOPS capacity, the user contents are manually moved onto a larger or faster (more spindles) volume, if possible.

The following figure illustrates a simple fixed-size provisioning environment using individual LU volumes on a host:

The alternative is to use a host-based Logical Volume Manager (LVM) when the planned workloads require either more space or IOPS capacity than the individual physical volumes can provide. LVM is the disk management feature available on UNIX-based operating systems, including Linux, that manages their logical volumes.

The following illustrates a fixed-size provisioning environment using LUNs in host-managed logical volumes:

In either case, hosts recognize the size as fixed regardless of the actual used size. Therefore, it is not necessary to expand the volume (LDEV) size in the future if the actual used size does not exceed the fixed size.

When such a logical volume runs out of space or IOPS capacity, you can replace it with one that was created with even more physical volumes then copy over all of the user data. In some cases, it is best to add a second logical volume then manually relocate just part of the existing data to redistribute the workload across two such volumes. These two logical volumes would be mapped to the server using separate host paths.