The three big driving factors that will impact the storage infrastructure over the next three years are going to be the increased percentage of virtualized servers within the environment, the continued adoption of solid state storage, and the potential for basically flat IT budgets. Virtualized servers increase the demand for performance, solid state storage has the potential to deliver that performance, and flat IT budgets mean you just can't throw hardware at it.
We discussed the importance of a performance specialist in a recent entry, but I think that the storage administrator who is planning for the future should consider that they should be that performance specialist. Being a performance specialist means understanding all the nuances of performance like latency, disk queuing, IOPS, etc. It also means mapping all of this performance data into an abstracted virtualized environment.
Unfortunately, there are not really any classes that we have seen that teach these things, so most of your study has to come from what is available online. Our "What is IOPS?" article and our "Visual SSD Guide" are good places to start and, of course, keep reading here as we continue to peel the performance onion.
Even with all the education in the world, you are going to need help. Investing in storage performance management tools is critical today. They should replace or at least augment the capacity management tools of the past. The key again is to be able to understand what these tools are telling you so when you see an event occur you know what the likely cause is and what you can do to fix it.
The second area to invest in is a thorough understanding of solid state storage. The first step in understanding SSD is to make sure you know what your performance needs are and where your performance bottlenecks are. That's basically the reason I put becoming a performance expert at the top of the list.
There are now at least four different solid state storage form factors competing for your attention and it may take a mixture of those solutions to achieve an optimal level of performance at an acceptable price point. Each vendor will claim their form factor is the best, but it really does depend on your environment, your performance needs, and your budget.
Another key to understanding SSD is not to believe the statements--or assumptions--made by most SSD manufacturers. The first step is to ignore all the 1 billion IOPS benchmarks that come out. Most data centers need nothing close to the extreme performance these systems claim to provide. Develop your own tests that better match your environment and, especially with SSD, perform those tests over a long period of time (at least 60 days).
Where should capacity management be in your future knowledge investment. Certainly capacity demands continue to increase, in some cases at alarming rates, but storage systems can scale to meet those demands now, and thanks to technologies like deduplication and compression the growth rate can be curtailed.
Capacity management is a process of investing in the right technology--performance management, however, is an art, one that needs skill and knowledge to be properly addressed. It is also an area where there is a shortage of skilled and experienced people that really know it. Where there is a shortage of skill, there is an opportunity for advancement.
Follow Storage Switzerland on Twitter