Let’s start with some fun facts:
- Did you know that banging your head against a wall burns 150 calories an hour?
- Did you know that SSD in servers is 8x more affordable than when it’s in an array?
- Did you know that on average CPUs are utilized only 25-28% of the time?
Superlatives in the technology space have numbed the market and made it difficult to see through the clutter to what is relevant. However, it is still worth trying when there are without question companies that should be heard. Datrium is one that deserves to be recognized for how they are fundamentally changing the data center.
Take a commodity server, load it with SSD, and turn that host and excess computing power, into a controller. The result has no effect on the compute performance of your VMs; and, in fact, IOPs are much higher because they don’t need to traverse the network to an array and back. Reads all take place locally, and writes are pushed to a low-cost shelf for redundancy. Plus, the system scales beautifully. Need more compute? Add a host! Unlike Hyper Converged you’re not burdened with lateral traffic.
Also unlike HC, if you lose a node, nobody cares. Your data is already protected in their netshelf. You simply pop on another X86 server and point the shelf to it. End-to-end encryption with data reduction to secure your data in-use, in-flight, and at-rest doesn’t hurt.
In real numbers, Datrium pumps out 20-40k IOPs per host with 8 hosts being a sweet spot. Using existing servers results in an all flash array that is half the price of contemporaries like PURE with 99.999% reliability. That’s 5x performance at 50% of the cost. It’s nuts!
Their biggest problems?
- VMs only
- It’s just not that sexy or easy to communicate the value proposition in a storage market that is dense with competition right now.
I really encourage you to take a look at their technology. Not only will they give you a free hat, but it is innovative technologies like Datrium that probably got you interested in IT in the first place.