I like SolidFire's technology and the way that it is implemented, from a node perspective instead of having a controller shelf architecture. One node can control everything, but if the node goes down, obviously the other nodes can bring everything back up. Going into the next generation data center, that's very compelling, as well as being able to use QoS settings and maintain a standard of performance for the VMs and things that are underlying it.
The SolidFire's technology and architecture allow for a more fluid and dynamic data center. It moves away from the controller and shelf design philosophy to a node design. This means that each node has the ability to control the entire cluster. In essence, you have the same number of controllers as the shelves.
If a node goes down, the other nodes easily take the load. This is accomplished both by the node technology as well as the Double Helix technology. If needed, you can easily remove one node and ship it to another location or attach it to a different cluster, with very little effort. The implementation of the structure is fairly easy, as well. Our first 5-node cluster from the box to serving data (for testing), took about five hours.
Improvements to My Organization
The biggest advantage is going to be the QoS settings, being able to maintain a level performance for our customers on whatever application that they're running at that particular time. For us, a business advantage is implementation time; our first cluster, four hours from un-boxing, racked, stacked and having it up and running.
We've had issues with the stability of our platform. We're a hosting provider and we've pushed it to its limits. We've found some of the bugs. The nice thing is that SolidFire has worked with us to correct those issues, bring new OS versions online to help correct whatever problems we've run into.
Scalability is excellent; it scales very easily. I can scale from a sizing perspective as well as an IOP perspective very easily. I can add a new node to the cluster. Within about an hour or an hour and a half, it's up and running. I have more space right then and there. Power and cooling is minimal as well.
Customer Service and Technical Support
The technical support is great. I've never had an issue with the technical support. When we open up a ticket – whether it's email, phone, whatever it happens to be – we usually get something back fairly quickly; they'll jump on the problem. We can give them access to the arrays or the clusters fairly easily so that they can figure out what the issues are. Getting new nodes or hard drives or whatever, in-house, usually happens fairly quickly. I have had a couple of issues with that in the past. Nothing major, but that's probably where they lost some of their points.
I was responsible for the initial setup in our Chicago data center; it was extremely easy. We had one of the SolidFire NetApp engineers with us when we did it and, as I’ve mentioned, it was four hours from opening the boxes to having it up and running.
Other Solutions Considered
We've evaluated three or four different all-flash type solutions. We actually went with one of the other solutions first. We ran into a very large bug about a year or year and a half ago, with an all-flash solution, and that particular provider was having issues correcting it. They still really haven't corrected it, so we can't push it as hard as we want to. When we were starting to look at a different solution, SolidFire came back with a good pricing model for us, as well as being able to meet the demands of what we're trying to deal with, and provide a rip and replace solution on our storage area that worked awhile back.
Evaluate everything. But if you're looking for an easy-to-implement solution from an all-flash perspective, really take a look at SolidFire. Try to get a PoC in house and run from a proof-of-concept perspective. The API's implementation is all very easy. You can look at it from your power and cooling aspects as well. That's the advice I have: Do a good proof of concept on the flash storage.