ILM and disk array virtualisation

One logical storage pool is the mantra of drive array virtualisation. It helps for easier migration between drive arrays and drives up disk utilisation. You can feed all sorts of disk into the pol: Fibre Channel, SCSI, SATA, etc. It all comes out in the wash as disk storage. A block is a block is a block no matter whether it is a Fibre Channel drive block, a SATA block or a SCSI block.

ILM is different. It matters exceedingly in the ILM world whether a disk block is a Fibre Channel block, a SCSI block or a SATA block; because you use that fact to decide where to put files. Fast and frequent access files go onto Fibre Channel blocks. Infrequently-accessed files go on the SATA blocks, presumably in some secondary storage tier.

How can you have the benefits of virtualisation - one logical pool - with the benefits of ILM - multiple storage tiers? A single logical pool of storage is, don't you think, a single tier of storage.

If that is the case then, for ILM to move forward in a world of drive array virtualisation, it's going to be necessary to have a virtualised pool of storage per ILM disk tier: The FC disk pool; the SCSI disk pool; and the SATA disk pool.

But, and also, in order to have a single point of management over the drive arrays and their pools it's going to be necessary to have an overall view of the virtualised pools. We can't have a server application asking locating disk blocks for a file it wants to access by asking each virtualised pool manager in turn ' do you have my file? Is it in the FC pool, the SCSI pool or the SATA pool?

Server applications need to ask 'the file system' where a file is. The file system, aka networked storage management facility, needs to know what different pools of virtualised storage it has and where files are allocated. The ILM facility needs to monitor file status in pools. When a file fails to meet criteria for staying in a particular pool then the ILM system tells the networked storage management facility that file X needs to move from pool A to pool B. Or it, the ILM facility, moves the file and tells the networked storage management facility that it's done so. Presumably all it needs to do is change file pointers in a table to reflect the new location.

This means, also, that there need to be virtualised storage managers for each separate pool of virtualised storage inside or underneath a supervising virtual storage manager, a manager of virtualised storage managers.

It's going to get complicated if ILM gets integrated with disk virtualisation.

A US supplier called Compellent has a virtualised SAN storage offering which can have two or more storage tiers inside it. 3PAR offers multiple tiers of storage inside a single facility too.

It seems to me to be irrefutable: in a virtualised disk storage world multiple tiers of disk storage mean multiple virtual pools under a single overall virtualised storage manager.