Conroe sits in the shadow of Houston, Texas, the de facto headquarters of the U.S. oil and natural energy industries. Our city is tiny compared our neighbor to the south, but like Houston we’re at the mercy of the oil and natural gas businesses. Being small only makes the ups and downs of those industries a lot harder to handle.
Until recently, Conroe’s IT department was like a hodgepodge of hardware and software bought when times are good and patched up when times are bad. Our aging servers hosted a mishmash of applications and almost weekly, someone was asking me to bring a new application online.
Demand threatened to overwhelm our antique servers and the buildings they lived in. Given how small we are – and how tight budgets tend to be even when times are good – we tried to make do with what we had unless it just stopped working.
A few years ago, I convinced our city administrator to invest in virtualization. I sold him by saying we’d need less hardware to get the job done.
The first phase went well: About 20 percent of our servers were virtualized. By February 2013, I was ready to do more. I wanted a new hardware cluster to support Conroe’s production application, database, and mail servers, and another to support vital operations during a disaster.
I used a VMware’s vSphere cluster in the first phase of the project. It consisted of a Hewlett-Packard 3000 chassis with four blade servers, and HP LeftHand storage. When I tried to move more of the production load onto that cluster, it became obvious that the HP storage system wasn’t up to the task. There was no way it could handle the performance needed by the new clusters. I had a dilemma on my hands: We’d standardized on HP. I asked myself: Could we afford to switch vendors at this stage of the game? I turned out that the better question was: Could we afford not to?
I considered other HP storage gear. At the same time, I hedged my bet by looking at other suppliers, one of them being Nimble Storage. The more I learned about Nimble, the more I questioned our commitment to HP. In the end, Nimble won me over with its combination of performance, efficiency, and elegance.
I liked how Nimble uses flash to maximize performance, but would allow us to store some data on cheaper hard disks. Nimble’s proprietary Cache Accelerated Sequential Layout (CASL™) file system compresses and coalesces data before writing it to those disks, making it much more efficient than other storage systems. And, Nimble’s solution would be easy to deploy and maintain, so much so that our existing staff could operate it with just a little training
Finally, there was Nimble’s overall cost – the price tag being only a piece of it. I wanted to know how much we’d have to spend to handle any future growth. It turns out that Nimble can grow incrementally. If we need more capacity, we add a shelf with more disks. If we want speed, we can upgrade just the flash, leaving the rest of the system alone. We buy only what we need.
And, Nimble’s scale-out capability allows us to cluster nodes giving us more flexibility than other storage solutions. That’s important since the nature of our economy makes it hard to predict what’s down the road.
Today, almost 90 percent of our production load is on Nimble and I couldn’t be happier. It does everything the company said it could do. And, the performance problems we had before are a thing of the past.