The endless variety of enterprise storage solutions means that you can find pretty much exactly what you need -- if you have your requirements straight. The trick to avoiding overspending or under-specification is knowing what questions to ask when you're deciding which option is right for you. Here are three questions I seldom see asked or answered with enough care or thought:
1. What is the expected lifetime of the solution I am designing?
If you want to prevent your budget from being shattered by unexpected expenses a few years down the road, give very serious thought to how long you want your storage solution to last. Sudden cost increases in the out years usually stem from unplanned capacity growth and unexpected late-year support costs. Define the period of time that your solution is likely to be used, and you reduce the risk of either eventuality.
In the data explosion, it's always tricky to assess your future storage without low-balling it or simply coming up with a fantastically large number. Giving yourself an accurate estimate will require lots of information about your storage needs over the past year or two, both the rate of capacity- and performance-load increases. Keeping future storage-intensive initiatives in mind, such as that pie-in-the-sky paperless office everybody keeps talking about, is important too. Either way, you can't make useful forward-looking estimates without knowing the period of time that you're planning for.
Other vendors scale their storage systems by adding entire shelves that include both disk and controller resources. This architecture tends to scale much more gracefully from a technical standpoint. As more disk is added, so is a comparable amount of controller and interconnect bandwidth is simultaneously, which essentially guarantees that you won't create a bottleneck. On the other hand, these architectures generally require more expensive growth iterations. If you simply need a small amount of additional storage, you might not be able to simply add a few disks; instead, you'd end up buying a whole new enclosure complete with controllers and a full stack of disks.
Regardless of the architecture you end up with, knowing how your solution will scale -- even if you don't think you're going to need to scale it -- is very important. Before you buy, ask yourself what the solutions you're evaluating would cost if you needed to scale them at twice the highest rate than you think you'll need to. Hopefully your estimates are accurate and you'll never need to do that, but the answer to that question can often shed light on critical scalability differences between the options you consider.
3. Will I have enough capacity to make use of the features I'm buying?
Capacity planning is an interesting process. On one hand, it seems to be fairly simple -- just add up the amount of storage you're using now and do your best to assess how much that number is likely to grow going forward into the future. But it certainly doesn't stop there.
One of the reasons we spend the kind of money we do on enterprise-class storage is that it generally comes with a lot of very helpful features such as site-to-site replication and block-level snapshots. You might even have the ability to de-duplicate your data directly on your primary storage array. None of these features come without costs: not just any licensing that might be required for them, but also less obvious costs related to capacity and performance.
For example, taking regular SAN-side snapshots of your data, a practice I highly recommend, will use additional storage space on your array that you may not have planned for. Depending upon the snapshot algorithm that's being used, your transactional disk performance may also take a hit. Estimating the performance degradation and planning additional performance capacity to combat it may save you from buying an expensive feature -- and turning it off because it slows your storage system down too much.
By the same token, if you're configuring your solution to replicate to another building or site and planning on using synchronous replication, you may need to plan additional disk and controller resources to handle the additional load that synchronous replication will exert on your hardware. The same applies to online de-duplication. Though a huge capacity-saving feature, online de-duplicaiton takes a significant amount of controller resources to accomplish. If that's a feature you're planning to use extensively, you need to know the impact it will have on your storage system's performance.
Good answers to these three questions will undoubtedly raise other, more specific ones. Coming up with detailed, realistic replies is vital -- and will give you a much better chance of buying a solution that will fit future storage needs, without a raft of unexpected expenditures.
This story, "Three questions for storage buyers," was originally published at InfoWorld.com. Read more of Matt Prigge's Information Overload blog and follow the latest developments in storage at InfoWorld.com.