Souped-up Coraid aims to change its spots
From pussycat to panther?
Comment Coraid used to be a downmarket beast, selling its Ethernet storage wares off a barrow on the street - now it's got shiny new management and has new aspirations. Can it make the transition and become a creature to be reckoned with?
Coraid sells storage using AoE, ATA-over-Ethernet, protocol which is very lightweight, making iSCSI look complicated and demanding of network resources. The company has been selling product through the channel with little marketing. It decided it needed more business "oomph" and the venture capital backers brought in new management.
CEO Kevin Brown, on a tour of Europe to boost the company's prospects with channel partners and the press, was operating in full boost mode. He said this about the situation he found when he arrived: "Coraid got to 1,000 customers in the Linux market with two part time support people… [the product] just worked… We came in and saw wet dynamite. Let's dry it out and light the fuse. The channel is hungry - they used to have alternatives like Compellent and 3PAR... No more. So there is a channel opportunity."
"We raised $35m through Menlo Ventures and others. [Now] we're now up to 1,400 customers… We have 130 universities as customers. We're getting pulled up into a lot of mid-range enterprises.
"Revenues in the last calendar quarter were five times higher than five quarters ago. 2010 revenues were nearly three times 2009 revenues. That's without concerted EMEA effort. We had Linux-based partners so it it was a foundation to grow on."
"We quintupled our ASP (average system price) since the new management team arrived 18 months ago… The company pre-us didn't have business people to take the company forward."
Brown and the Coraid backers see an opportunity to slip the thin end of a low cost AoE wedge into the storage array market and capture a lot of low-end and maybe mid-range business. They are hammering away at this wedge for all they are worth - and they will be worth a whole lot more if Coraid succeeds.
Brown sees a transition opportunity opening up in storage because of mainstream storage vendor pricing; he refers to these vendors as "the cartel".
"Change has to happen in storage because storage costs too much. There's a fundamental mismatch between raw HDD costs and the price points the cartel is offering. There is a generational shift from Fibre Channel (FC) to Ethernet, from big iron to commodity scale-out."
"We're bringing in flat Layer 2 Ethernet, [with] jumbo frames, and take full advantage of the switches. Ten gig ports are dropping below $500… We start at less than $600/TB for our SRX line… and [with] 4U and 36 drives our price/performance story is a 5-10X advantage over FC arrays. We're faster than FC and about 1/5th price. It's typically $3,000-$10,000/TB for a FC array and we're starting at under 1/5th of that."
"We can set up an array in 60 seconds; three commands; it's done. After that it's self service in the hypervisor. You configure storage as if it's the C: drive… With us multi-pathing goes away, port zoning and bonding goes away. The whole idea of a fixed connection network goes away.
"Instead of building reliability into the pipes, which has a lot of complexity, we build reliable delivery into the initiator and the target. iSCSI tried this with TCP/IP - but that's email and latency is 200 milliseconds round trip. It's crazy protocol to use to send data 1 metre.
"We use different and better code. Our round trip latency is measured in microseconds; it's bare metal, raw L2 Ethernet frames."
"We're working with Arista. Andy Becholsheim is not exactly a fan of FCOE (Fibre Channel over Ethernet) - it brings a whole later of rigidity onto something elegant and flexible that's Layer 2 Ethernet. We're massively parallel - it's datagram-based."
He's banging away at the cost, simplicity and speed drums, hoping to drown out competing marketing messages from mainstream suppliers. Until we have independent benchmark numbers we can't actually verify what he's saying about Coraid's product performance in an objective way.
What about the accusation that AoE is proprietary? Brown said: "It's been native in the Linux kernel since 2005. Other companies are using it, for example, Whiptail. The existing trad array suppliers depend on FC-style networking. They could slap AoE on the side. Our competitive assumption is they will.
"The AoE potocol is really an un-protocol; it's 12-pages long [and] we've posted it on our website. The iSCSI protocol is 100s and 100s of pages long, ditto FC. We have taken that complexity out of the network."
But it may be that the reason that the iSCSI and FC protocols are so detailed is that they do a whole lot more and the resulting network connectivity is more reliable and robust. A bicycle is a heck of a lot simpler than a car but the two machines are meant to do different things and can't be substituted for each other.
How would he contrast Coraid with Nexsan, Pillar Data and Xiotech? "They all did vitamins, storage supplements, clever software to make storage arrays go faster." Hs notion is that they overcomplicated things when what was needed was more simplicity, and that's what AoE brings.
Brown presents Coraid as having a fire in its belly. "Our deal cycle is 35 days. Our win rate is very good‚ the starting cost is $20K or $30K. Customers find a small project. We land and then we expand… Everyone's pretty excited."
We mustn't forget that Brown is pumping up the prospects of his company. He's a CEO. If he didn't do this stuff then he doesn't deserve to be in the job.
Coraid is setting up an expanded and locally managed EMEA channel. If Coraid, its boosting CEO, and its channel can convince customers that Coraid's cost and simplicity advantages are real and that its storage functionalities are robust, reliable and dependable, then it will have turned a corner and can start growing strongly. ®