Under what circumstances should an IT organization consider the purchase of an all-flash storage array?
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
This is a little glib, but I would always buy an all-flash array for everything if I had more money than God, and therefore I could always get the latest, greatest thing because -- well, why not? The only reason we put data anywhere outside of memory is because we can't afford to put it all in memory. Or, maybe if you are in the sort of business where you want to impress your customers and clients that you have your entire data center in one rack, and it goes faster than the speed of light, then absolutely do that.
For more normal people and the vast majority of users, I don't think they should buy an all-flash array because it's an all-flash array. In other words, don't get attracted to new, shiny things just because they're new, shiny things. The first and foremost thing, as with the purchase of anything in life -- but particularly in IT -- is that you need to start with what you are trying to do and determine whether the devices that you're choosing between do at least what you need.
Next are a couple of things to consider especially. One is functionality. Clearly if you are expecting fairly rich functionality with snapshots and thin provisioning and yadda yadda, then you need to ensure that that is within the all-flash device or can be used with the device that you're looking at. The other thing, from the shiny object side, is that it's all too easy to get carried away with chasing specifications and end up buying more than you need.
You certainly need to be looking at the capacity, scalability and the ease of use of the system. There's no point buying something that can't grow with you. And there's no point buying something that is so difficult to use that it actually wastes more manpower than you gain somewhere else.
And perhaps most important of all is the financial side. I hold to the assertion that every purchase in storage is a financial purchase. How are you going to use this device? Are you going to use it for persistent data with certain "point" applications, or maybe in some situations -- where you can both afford and need to do it -- put your entire infrastructure on this flash unit? Or are you going to use it as a caching or tiering device? That will obviously impact both the amount you need to buy, and more importantly, it will impact the financial implications.
The judicious use of some amount of flash can preclude a lot of wasted spinning-disk capacity that otherwise has traditionally been bought purely to get the IOPS and not the terabytes. The overall economics of some amount of flash can actually be very positive on your overall cost per gigabyte.
I don't think there are particular applications that are silver bullets for all-flash arrays. Clearly the sorts of features that have higher I/O requirements and high read requirements, such as databases and virtual desktop infrastructure, are good examples, but they are by no means the only ones.
This expert response is part of an interview with Peters conducted by SearchSolidStateStorage.com Senior Writer Carol Sliwa.
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.