Ethernet SAN vendor Coraid Inc. changed CEOs today, as Kevin Brown, who stepped aside after more than three years on the job, was replaced by former NetApp executive Dave Kresse.
Brown will remain on Coraid's board and said he plans to take an active role in setting company strategy and working on industry initiatives such as OpenStack. His replacement, Kresse, was formerly NetApp's vice president of storage management from 2003 to 2007, and was CEO of security testing vendor Mu Dynamics, which was acquired by Spirent in 2012.
During his term at Coraid, Brown expanded the vendor's EtherDrive products that use the ATA over Ethernet (AoE) protocol -- an alternative to iSCSI -- to include network-attached storage virtualization, solid-state drives and a cloud storage platform. He said he expects those initiatives to continue under Kresse in what the two described as a highly disruptive time for enterprise storage.
We spoke with Brown and Kresse about the CEO change and their plans for Coraid.
What was the reason for this change?
Kevin Brown: I'm moving to the board of directors [and] I will be quite involved in helping out. Dave and I have known each other since we were vice presidents at NetApp years ago. I'm going to stay involved in all of the strategy and company initiatives like OpenStack. Dave will drive our scaling strategy and we'll work together to fill out all the platforms that put us in the enterprise.
Dave, how close have you followed Coraid? What prompted you to take this job?
Dave Kresse: A couple of things stand out. The first thing is the market opportunity the company has. You have a $30-billion-plus storage market that has lots of innovations going on around it, like flash, but I think there is a multi-year transformation happening in the data center. We're moving from an environment with small data running on large storage systems toward environments that start to resemble what's happening on the compute level, where you have large data sets running across lots of small boxes. You have to think in terms of scale-out. And if you look at the strategic vision that Kevin has put in place, I believe the market is [changing in our favor]. We have the opportunity to become the right storage for these environments.
Why make this change now?
Kresse: The time when a change like this makes sense is when things are going smoothly and the change can be non-disruptive to the organization. Kevin's involvement in the company will ensure that continuity.
How can a company like Coraid survive against the big storage players?
Kresse: What we do to survive is become the best solution for the class of problems that the incumbents aren't focused on. A lot of innovation in the storage industry is still around incremental improvements in traditional scale-up architectures. We have taken a different approach in terms of the problems we want to solve; we're looking at the world of the data center transforming into more of a scale-out architecture. By taking on something that is a different way than how the data center works, and being the best at it, is how you build a long-term, independent company.
I think the endgame is around the software-defined data center, but there are steps down that path. EMC is just starting to make noise around projects that look at scale-out.
Brown: The type of scale-out and automation that already happened in compute and networking is a tsunami rushing to overturn storage. Traditional platforms get refreshes and vitamins to look more modern, but there are big architectural changes coming. We're moving to something more scale-out and agile, and it will run on Ethernet, not Fibre Channel.
The big guys are starting to talk about solving the problem -- EMC is shipping ViPR soon, but that's taking silos of complicated storage and trying to tie them together.
There are a lot of iSCSI storage systems out there that run on Ethernet. Why haven't other vendors gone to the AoE protocol that you use?
Brown: A lot of storage vendors have very complex operating systems and management systems tied around those [iSCSI and Fibre Channel] protocols. They haven't seen a need to change because they were pushing toward Fibre Channel over Ethernet [FCoE]. In the last year, that has faltered as a strategy. So their big strategy for getting off iSCSI and getting to a high-performance, low-latency Ethernet has now crumbled. They thought they had it covered with FCoE and they don't. So as we move to a world with increasing performance requirements with flash, moving data over Layer 3 iSCSI is increasingly looking inadequate. It hasn't gotten the uptake for high-performance enterprise workloads.
So I take it you see no need to move into Fibre Channel?
Brown: Our customers are not asking for that; they're actively looking to get off of it. We have 10 Gigabit Ethernet [GbE] today and we use it massively parallel. We can deliver 10 GbE, 20 GbE, 30 GbE and 40 GbE just by plugging in more cables. We're already at a much higher performance level than current Fibre Channel. And 40 GbE is around the corner. We think the next couple of years is where it gets super obvious that if you want to stay on the performance curve, it ain't gonna be on that old mainframe network.
How do you define software-defined storage?
Kresse: You have intelligent software that programmatically manages services on top of commodity hardware. That, to me, is what it is. And that commodity hardware has to almost act like Lego blocks where you can just scale it out seamlessly and not put a big operating burden around that. These environments are scaling at unprecedented rates and the need is there for companies to add capacity on demand. Companies like Amazon have shown a clear blueprint of how things can work and how they should work. You now have people building their own data centers, looking at that in terms of the way it should feel for end users to rapidly spin up new services and workloads, as well as the economics of that.
Will you need a virtual appliance to fulfill that vision?
Brown: Our management system is a virtual appliance. We use commodity, off-the-shelf hardware for our storage array, but we have a tested stack that customers like to procure from us. We've gotten there ourselves on commodity hardware. All of the magic is in software.
How important is flash to the Coraid strategy?
Kresse: Flash is a component. It's something we work with and our customers utilize, but it's an orthogonal question to what we're really focused on, and that's the scale-out aspect. But for me, it's not a transformational technology. It's another medium for applications to run over. But the change from small data running on large boxes to large data sets running on lots of small boxes is a fundamental reconfiguration of the data center.
Brown: Flash is super disruptive, but not an industry in itself. It's a drive type, and everyone is investing heavily to take advantage of it, including us.
What's your cloud strategy?
Kresse: Cloud is a central part of our strategy going forward. We see the market shifting to cloud efforts on the public side with service providers as well as private clouds being built. We have some very big customers in that space and we partner with these folks at the cutting edge of building these clouds. They run into challenges and we partner with them to help solve them. Then we get that expertise and can turn around and help other customers who are right behind them.