As discussed in my last blog, many storage vendors talk about storage tiering, but they often mean “in the box” tiering that supports a variety of drive types within a single array.  Considering that most data only stays active for about 30 days after it is created, it makes sense to migrate that data to denser, slower disk drives to save money.  But if the data was created in a tier 1 storage system, there is a high price associated with buying the system no matter what types of drives are included – the system was designed from the ground up to meet demanding performance needs.  Moving data within one of these systems from high-performance storage – disk or solid state – to slower denser drives still carries the “array tax” that comes with the high end enclosure, electronics and operating system.  Users need a way to do cross-platform tiering to get less active data (about 80% or more of the data in the data center) off of the expensive tier 1 systems and onto systems that don’t have a Ferrari engine at their core.

Today’s NAS environments seem to have a jump on this one.  One reason for this is that NAS systems have knowledge about the data.  They know when a file was created and when it was last accessed.  They can identify, and migrate, the 80% or more of files that haven’t been accessed lately.   Another reason for the NAS jump is that the data management layer in some NAS system is software-based, virtualizing the underlying hardware environment.  That means users can transparently move data to a lower cost system (disks, enclosure, electronics and all).  A number of file systems have built-in policy engines that automate the data movement so it can be operationalized.

A challenge for block-based systems (and a number of file-based systems) is the tightly coupled relationship between the storage controller and the disk drives.   These systems need a higher-level application or appliance to help move data from tier 1 to bulk storage tiers, typically an archiving application or virtualization appliance.  If you don’t have some type of higher level abstraction that manages communication and movement between platforms, you are likely paying a “tiering tax” on all the data stored on the slow, dense drives that are surrounded by high end controllers and electronics.

There are “out of the box” tiering solutions available for both block and file data.  Both F5 and AutoVirt offer file system virtualization solutions (AutoVirt is Windows-only) and a number of companies offer block virtualization including HDS, IBM and HP.  There are also some innovative solutions coming into the market, like those from Avere and StorSpeed, that take a new approach to the problem and “cache” performance sensitive data while storing everything else in 3^rd party bulk storage – both of these solutions are targeted at NAS environments and take very different approaches, but both show what can be done when you don’t start with an array-based approach.

So what should end-users do?  Like the answer to almost everything in the data center, it depends.  Not every data center is the same so there is no prescriptive response.  But, users can ask their storage vendors, push their storage vendors, to provide solutions that help migrate data between physical tiers.

Related posts:

1. Why ILM Never Really Took Off

Tags: AutoVirt, Avere, F5, ILM, StorSpeed, tiering