Coraid CEO Kevin Brown.
You’ve had a number of roles I see, but were also one of the founding team at Inktomi. Tell us a bit about your background and how you ended up at Coraid.
It’s funny, I was an economics major at Berkeley. I did economics and finance research. We were quant jocks, so we were programming in Fox Pro, using the mainframe at Stanford. We were geeks, technology users. This is 1990. I got my first copy of the Mosaic browser in ’93, a beta copy, and we just freaked. We knew this was just gonna change everything. Democratise technology. So I went back to grad school at Berkeley, and we built one of the very first advertising engines, before Netgravity or DoubleClick or any of those.
We ended up merging that technology with the Inktomi founders, so I joined as the first employee after the technology founders. I was there 7 years, vice president and general manager; and we went from zero to $200m revenue in five years. We powered MSN, Yahoo, AOL – we were number one or number two in search for a decade. We were running on SPARC, you know Sun, and we saw Google come in riding on x86, and they had a 5x advantage, which made all the difference. So that lesson wasn’t lost on me.
So if you look at the storage industry today, it looks like SGI, DEC, Sun – these expensive, complicated midrange things. Where people are going with this massive data is that they need a big pot of Lego-like building blocks. Much less expensive, much more flexible. I saw the same pattern match in the server business, that I saw with the Web, and I saw the need for it now in storage. Here’s a company [Coraid] below the radar that went from zero to a thousand customers with no funding, no sales and no marketing. Some of them Petabyte scale, so the National Institute of Health in the US, they run the human genome project on Coraid storage. IBM Blue Gene uses Coraid. Harvard, MIT, NASA.
I looked at 130 companies and I went, ‘oh my God’. This has the chance to come in and be a real disruptor and handle this exploding data, where the existing technologies are going to hit a bottleneck.
Have you raised funding since?
In the last two years we’ve raised $85m in funding in a pretty tough environment. We’ve multiplied our sales by 7x in two years. We were already doing millions and now we’re in the tens of millions. We’ve now passed 1,500 customers. We’re hitting that expansion phase. We have a good base in the UK and the rest of Europe too – there are a lot of information intensive businesses, from pharmaceutical research to video surveillance to creative web businesses.
What impact do you see the down economy having on storage demands?
People’s storage is growing 50% per year or more, but their budgets are flat. So that’s a great opportunity for us because we’re coming in with something that’s faster than what they have and simpler, at one third to one fifth of the cost. In many cases we take projects that weren’t viable and we come in and put together an architecture.
Speaking of architecture, tell us about the Coraid technology itself.
Some of it was working out how to use raw Ethernet instead of this expensive storage networking.
Is this where the acronym AOE comes from?
Yes, ATA [a storage protocol] over Ethernet, a piece of open source technology that we contributed to in the Linux community. It’s been great because the economics of Ethernet has changed – 10Gbit has now dropped below $500 per port. It used to be $3,000. So you have this wave of very inexpensive networking, and none of the existing storage vendors can really take advantage of it. We’re the only one that’s native.
But haven’t a lot of larger companies made a big investment in fibre channel? Is this not quite a scary move for them?
Well a lot of them have either hit a roadblock or they are having to look at upgrading to the next speed. Every time there’s a forklift upgrade. Or when people have a maintenance renewal we are able to say for the price of your maintenance renewal we’ll give you double the capacity that you had before. The storage vendors had premium priced everything and it’s custom hardware. We’re able to come in with off-the-shelf x86 hardware.
It fits well because you used to have the mainframe and Unix boxes but now you have x86 pizza boxes. With virtualisation it’s a very elastic resource. Then you connect it to a storage network which is very rigid, and to a big box that looks like a mainframe for your storage, which is crazy. You’ve got all this great economics at the computing layer but the network and the storage layers still look like the mainframe days.
Do you have virtualisation in the box?
Yes, we can do storage virtualisation, very small or very large LUNs [logical units]. It’s one of the things we changed to move up-market into big enterprises and participate in the cloud market. We built that in in the last two years.
Do you have thin provisioning?
We have most of the pieces of that today with our logical volume management, but we’re adding the thin provisioning feature later.
Roughly when, when you say later? This year?
Does it have a codename?
We’ve got a couple of release vehicles and we’re deciding what’s on the train. TBD.
How many in engineering now?
The board has asked us to stay vague on that but the overall company is in triple digits.
Over 100 but less than 500?
Mostly in the US with an office here?
Yes we have a small office in London, and office in Australia. Resellers throughout Europe. We did our fifth deal in Nepal. We’re predominantly through the channel, apart from some cloud customers who buy direct.
Gartner has said the ‘personal cloud’ will replace the PC at work by 2014. That’s a big change.
It’s a big change. You’ve seen it with salesforce.com, different application areas.
Yes, I like the story of Workday, the on-demand HR company co-founded by the PeopleSoft guy, Dave Duffield. They were called in to talk to a company that was thinking about becoming a customer, and they found out Workday was already being used in their company, just under IT’s radar. Business units had just started buying it on their credit cards.
We get that sort of thing where you do a small virtualisation deal and then they are like, ‘we want to talk to you about some of our other projects too’. We may not be their biggest vendor to start out with but once we have done a small project of course we are in a very credible position.
What about the brand? You don’t have the brand cache yet of someone like an EMC.
That’s part of how we have to qualify deals because if someone is just sticking with the big guys, we figure that out pretty quickly. But there are many people that have real problems to solve and they are not getting solved by their big vendors. And there are a lot of people that actually feel pretty badly treated with tough pricing and aggressive sales. End users have gotten the squeeze. Storage has gone from 15% of your budget to 40%.
As well as the EtherDrive product which is disk, you have EtherFlash too, a flash-based array. Can they sit on the same network?
Yeagh, you can mix and match down to the drive level. Think of EtherFlash as a bundle that adds flash to EtherDrive. What you get is high performance flash drives with the scale out architecture of EtherDrive. You also get nearly 200,000 I/Os per second for under $10 per Gigabyte. You can get up to 200 Terabytes in a rack.
I met one of Whiptail’s co-founders recently. They seem to be doing some interesting stuff with all-flash storage.
They’re good guys, smart guys. We view a lot of the flash innovations as a micro innovation in terms of, ‘how do I get value from this new drive?’ We think we have a macro innovation which is around how you change the architecture of the data centre. How do you handle this massive wave of data expansion and simplify and automate this? We’ve just bought a company [Yunteq] that does cloud automation, because you have Petabytes and Petabytes of storage and you need better tools to manage that. Yunteq was the guy who was the former CTO of networking for Sun, and guys out of Akamai and Kashflow – very experienced engineers.
What will the Yunteq acquisition add to your portfolio specifically?
We’ll add a lot of automation, policy-based provisioning.
When will we see the fruits of that work?
We’ll be launching some of those products this year.
You already have EtherDrive and EtherFlash, should I assume there will be an EtherCloud?
[Laughs] That may already be taken. It’s TBD. We’re not like Apple that just trademarks every word that starts with ‘i’.
Thanks very much for talking to CBR today.
Thanks for your time.
Please follow this author on twitter: www.twitter.com/jasonstamper
This article is from the CBROnline archive: some formatting and images may not be present.