Storage

Disk storage systems: Purchase fully populated or not?

There is no more uncomfortable situation than to be out of storage for your infrastructure. IT pro Rick Vanover shares perspectives on whether or not to purchase disk systems fully populated.

Over the past year or so, I've been exposed to a number of events and people that are giving me insight in to storage technologies. While my historical background is not storage administration, I am quickly becoming a storage expert from my current activities as a blogger and my professional role. One person I've been following recently in storage circles is Chris M. Evans, also known as The Storage Architect on his popular blog.

Recently, Chris and I had a discussion at the launch of the Virtualized Storage Platform (VSP) by Hitachi Data Systems (HDS). At the event, we were discussing purchasing patterns for storage. Chris practices more in the enterprise storage space, where I am in the modular storage arena; but we found common ground in the discussion on how to purchase storage. Chris' recommendation is to purchase a storage system fully populated to the maximum configuration, regardless of product. My practice has been to purchase modular controllers and small inventory of drive shelves, leaving room for expansion. The modular storage I work on usually has a maximum supported raw storage capacity of around 150 TB.

There are pros and cons for each purchasing pattern. In the case of purchasing modular storage systems, adding additional shelves is relatively easy and affordable. The initial investment of the storage processor is usually the hardest part of adding the modular storage system, but is an easy way to have granular control of what types of storage reside on each shelf. For example, it is easy to add a shelf of SATA storage, then turn around and add a shelf for high-speed SAS storage for a different requirement. Further, it can be difficult to forecast the storage consumption making a pay-as-you-go (rounded up to the nearest disk shelf) approach an attractive strategy.

In the case of purchasing a unit fully provisioned, Chris illustrated that the storage administrator is given ultimate flexibility. If there are hot spots on existing storage, the offending logical unit number (LUN) can be moved to a quiet area on the fully-provisioned storage system. This benefit can work for very large storage processors like the VSP, but also could work in modular storage processors as well. If the storage processor has support to migrate a LUN from one aggregated collection of disk to another, the hotspot can be avoided there as well.

The only downside of purchasing a unit fully provisioned would be if a different tier is needed in the case of SAS or fibre channel drives versus SATA drives. One strategy is to fully provision the new storage system at a distribution that is consistent with your forecasted needs. For example, if you currently have 25% faster storage (such as SAS or fibre channel drives) and 75% slower storage on SATA; the new purchase can provision the tiers to that distribution.

One could make a number of purchasing strategy recommendations based on any number of requirements and factors, but the central question is do you purchase storage units fully populated or not? Share your comments below.

About

Rick Vanover is a software strategy specialist for Veeam Software, based in Columbus, Ohio. Rick has years of IT experience and focuses on virtualization, Windows-based server administration, and system hardware.

Editor's Picks