Virtual storage capacity management: an admin’s worst nightmare?

Most hear “server virtualization” and think: efficiency, ease of management, high availability and flexibility. But these benefits – the aim of sound IT planning – really only extend to the server (and in some cases application) layer. Administration, it turns out, is a whole other kettle of fish.

That’s because the complexities of introducing server virtualization into an environment force administrators to spend far more time than in the past on planning the overall capacity requirements of an environment and how to lay down data to ensure that the benefits virtualization brings to servers isn’t offset by problems to the storage environment.

Here are the three most common technology features created to help alleviate this pain point – as well as some of their pitfalls:

Thin Provisioning: Thin Provisioning allows administrators to show an OS/App/Hyper-visor an amount of storage they can grow into, while not actually allocating the physical space on the SAN/NAS. The SAN/NAS only allocates storage as data is written to it, so administrators can spend far less time planning — and only need to purchase and allocate what’s needed NOW, versus in 6 or 12 months.

DWhile thin provisioning provides a lot of value – extending existing capacity lifespan and lessening the number of tasks to manage virtual machines and data — it also causes issues. Administrators need to watch storage utilization very closely for any sudden explosive data growth that can lead to storage suddenly filling up and reaching its capacity maximum. That can stop growth and crash systems.

While it can help alleviate many bothersome tasks, you need to be careful with setting up thin provisioning to ensure there are proper rules to monitor capacity levels so there’s always room for growth while still ensuring storage use is as efficient as possible.

Primary Storage Data De-Duplication: More controversial within the IT community as a viable solution to capacity management in virtual environments, Data De-duplication uses algorithms to analyze data and find like “blocks” between different data streams or data sets. When it finds a duplicate block of data it simply leaves a pointer in that location to where the original block of data can be found. Removing duplicate blocks can allow for massive capacity reduction.

One area where de-duplication can be counterproductive is with long term capacity planning. When data is not de-duplicated, 1MB of written data = 1MB of deleted data. With de-duplicated data sometimes you’ll erase 100s of GBs of data, and because specific files/blocks are just “pointers” to other data that can’t be erased, you don’t actually delete much, if any, space from storage because of these complex dependencies.

Data Tiering: Data tiering software moves blocks of data in a way to allow you to purchase tiers of disks for your array (SSD, FC, SAS, MDL, SATA) so you can place data with lower value and performance requirements on larger capacity, lower performing, lower cost disks, and mission critical, performance-oriented data a small amount of performance-oriented SSD disks.

Tiering allows you to spend less and ensures all drives are used at all times, driving up utilization and efficiency.

This sounds like the greatest thing since sliced bread, but there are pitfalls, of course.  One for instance is that if data is moved in very large increments (1GB) a lot is moving through the SAN/NAS heads, when all that may have been needed was 256KB. So each time a block is moved a lot of extra data follows it, meaning the expensive tier1 disk pool may need to be larger than in a system that moves data in smaller amounts. Another consideration is the types of drives: if the only tiers being used are 7200 RPM SATA HDs and lightning fast SSD, the gap may be so large between the two that tiering won’t be efficient.

While there are certainly challenges, these technologies can have a real positive impact on administration. Just ensure there’s careful planning and discussion at each step of the process and you’ll likely get your desired outcome of an efficient, flexible and virtual ready storage environment.

Related Posts:

One console does it all – and cuts admin time by up to 90%

The skinny on Thin Provisioning

Related Posts

Isn’t It Time You Hopped on the Virtualization Bandwagon? – UPDATED Editor's note: our virtualization expert Stephen Akuffo weighs in on the Virtual Space Race Study blog series - see his notes below. You’ve probably researched how virt...
RECO Gains Peace of Mind with Cloud-Ready Storage The Real Estate Council of Ontario (RECO) is a not-for-profit corporation responsible for administering the Real Estate and Business Brokers Act, 2002 as well as associate...
Why Storage Tiering is Relevant to You There are a lot of differing opinions floating around these days about “Storage Tiering” and the different methods of achieving it. Before we get into some of the popular ...

About Adam Wolfson

Adam Wolfson is now an HP Technical Architect here at Softchoice and holds more than 8 enterprise technical architecture certifications from HP.