Brian Morin: As you know, virtualization is sweeping the landscape. Gartner has estimated 74% of workloads are virtualized right now. I think IDC is putting their forecast at 71%. Gartner has mentioned that by the end of this year, 80% of workloads will be virtualized. As a lot of companies are virtualizing their most I/O-intensive applications and mission critical workloads, their storage back-end is not able to keep up with the new performance requirement.
They all go into this virtualization process blindly regarding what they need on the back-end storage infrastructure to support it. They have a very difficult time scaling that virtualized infrastructure because they reach an I/O ceiling on their back-end storage. They run into an I/O ceiling and have to stop scaling their virtualized infrastructure. For our customers, they’re almost in a death spiral of constantly having to overbuy on more storage just to keep up with the SLA of their application performance.
That’s typically when they find us because they’re now looking at doing a very expensive back-end hardware forklift upgrade before it was in budget. They find us hoping that we can bring a solution for them and are amazed to find out that just by using our software, they’re able to solve their toughest application performance issue and stay within budget.
Sramana Mitra: What are the top trends in your segment of the industry?
Brian Morin: We’re in the I/O performance space. If you look at the storage market right now, you’d agree that the storage market itself is more fragmented now than it’s ever been in the last 20 years. There’s convergence, hyper convergence, on premise, hybrid, or 100% in the cloud. Even the storage arrays themselves with the spindles or flash.
One trend though that we see in front of us is hyper convergence. Hyper convergence is decoupling storage performance from the architecture and moving that performance up to the compute layer to get as close to the processor as possible. The trend we’re seeing is that performance is moving closer and closer to the compute layer because so many applications are critically IO-intensive.