• Home

Exclusive Interview With Brantley Coile, CEO and Founder, Coraid

Company back to ATA-over-Ethernet thanks to its tenacity
By Philippe Nicolas on 2017.08.02

AddThis Social Bookmark Button




Who is Brantley Coile?

He is:

He was:

  • CEO and founder, The Brantley Coile Company
  • Founder, chief scientist, CTO and board member, Coraid
  • Development manager, Cisco
  • Founder and CTO, Network Translation (acquired by Cisco)
  • Senior software developer, Adaptive
  • Member of technical staff, Bell Labs
  • Progammer, DTS, and
  • before that at University of Georgia as manager systems group

StorageNewsletter: Could you illustrate your career path until the start of Coraid?
Brantley Coile: I'm an old UNIX guy from the 1970's who just never stopped coding. I also grew up in an entrepreneurial family. We were running our own businesses for as long as I can remember. I suspect I learned more from my mother growing up than most learn getting their MBA. As far as Unix goes, I started with sixth and seventh edition UNIX.
Later, I worked for various telecom equipment companies, embedding seventh edition Unix into their equipment that needed a better operating system. Embedding Unix in stuff, was swimming against conventional wisdom of the day. I got real used to doing what made sense to me and ignoring what others said was the way to do things.
In the early 1990's I left Silicon Valley and returned to Athens Georgia, where I co-founded Network Translation Inc. with a buddy back in Palo Alto. I developed network address translation (NAT), stateful packet inspection, and virtual private networking for our new product, the PIX Firewall. We sold that company to Cisco who in turn sold about $4 billion worth of the product. Then, while at Cisco, I invented web load balancing. That product was the LocalDirector and its sales was about $3 billion. Cisco did okay off me.
For the past 17 years I've been working in high performance, low cost network storage. That's when I designed the ATA-over-Ethernet (AoE) protocol.

Back to the genesis of Coraid, what did you anticipate and why did you create a new storage protocol? Did the market need this?
I look for shifts in the IT world and ask how can I make something simpler. What's changed that makes the current conventual wisdom obsolete? It takes understanding why things are the way they are and asking if those reasons still exists. This, I learned from my dad.
In the late 1990's I saw the need for a new network storage protocol because of a couple of things. First, Ethernet was trending way ahead of FC in speed. It was slower but you could see that it was going to be faster. The Ethernet volumes were huge which meant faster technology development and lower product cost.
Second, once you were thinking of Ethernet you realized the per port cost was so low you could use more than a single port in each server and storage appliance to scale performance and improve robustness. This meant you can't use the concept of a connection as we know it. So TCP was out. We then looked at what was the simplest thing that would do what you need to do for network storage, IO, discovery, fencing, and later, reservations. We spent five years getting AoE right. It has proven to be very useful. Very robust to changes in the media.

Could you describe the advantages of AoE vs. other classic storage protocol such iSCSI or FC?
Gee. Where do I start! One thing it does is make the user feel like a genius. Anyone can fully understand the technology. The spec is only six pages.
As to the old stuff, FCl is the oldest so let me start there. It's in essence the descendent of the IBM ESCON fiber optic channel of the big mainframe age. Very complex so one can hand craft configurations for optimal results. It assumes you have a lot of money for people and a lot of money for equipment. It assumes things are pretty static. For example, the way they do flow control is called queue depth and it's set by hand. Add a new server, go in and change the queue depth. Works well for the big iron boys, but not so much for everyone else.
The other popular protocol, iSCSI, is an Internet Engineering Task Force protocol that was meant to put FC like services on the Internet. That's what the "i" stands for, Internet. No one used it that way, though. The variations in round trip times freaked out the file system layers above. Besides, there are better ways of moving data streams across the Internet. Both of these technologies assume a single port per system, so doing multiple paths and load balancing requires more levels of complexity. AoE was designed to run directly on Ethernet over multiple paths via multiple ports on each box. It uses dynamic configuration and congestion control at three orders of magnitude faster time base than TCP. IO requests flow through the network like a fine wine. The fact that the specification is only six pages and easy to understand means the user can sleep well knowing how things work.

During the previous life of Coraid, you offered block and file storage appliances? Any quick details on them? And what about the price?
Our SR products sold very well. The cost was around $4,000 per 15 bay unit. You add your own drives. The performance was great and just got faster. It was cast iron solid, to. We have a lot of people who are still using the systems. Some have been running them for over six year without rebooting. We also sold a pre-configured Linux system to let people use the storage through NFS.

At a normal pace, what was the revenue of Coraid? And the growth rate?
I divide Coraid's history into three phases, sort of like the life cycle of a butterfly. First, was the Classic Coraid, when we first started. We went from zero sales in 2004 to $12 million in accumulative by 2009. We were profitable from the start. The next phase was the California Coraid phase after we started taking venture capital so we could do a faster job of letting people know about now great we worked with VMware. Sales peaked at a run rate of $48 million in the last quarter of 2012. A total of $100 million in sales of AoE appliances were sold. Now we are the new Coraid, just getting started again. We are back to being profitable. I now own the company. We are here to stay.

What happened to Coraid? How did this company disappear in just few months? What are the root causes? Could you share with us some info on that and your feelings as the Coraid founder?
When I accepted venture capital, I knew we were going to be on track to go public. That meant putting the focus growing revenue, raw sales, and to not focus too much on profits. What I didn't know was the huge losses the VCs were willing to take to grow sales. It puts a company in a position of relying on more and more investment to cover the cash burn. It was a lot more burn than I thought going in.
But this scheme only works if you keep sales growing, like they did from 2010 through 2012. But then management made a huge mistake: they decided to change the operating systems. Instead of using the Bell Labs technology I had used, and still use, they wanted to switch to Solaris. Companies like Coraid can't afford to change OSes no matter what the reason. It confuses customers. It means completely changing developers. It stops new features as the new team relearns all the lessons the old team had already discovered. Sales sagged. Funding disappeared. With a high cash burn the end comes quickly.
It hards on the founder. It helped that the I liked the team, even when I disagreed with them on this change. Everyone acted in good faith for the best interest of the company. But I got a virtual PhD in sales and marketing. And I learned a very important lesson: if it's your vision, never give up control.

We understand the failure was not related to technology and products except the change to Solaris, but why other companies did not try to acquire some asset from Coraid?
There was interest in the technology. A couple of bidders showed up but didn't bid enough for the venture debt that foreclosed on the company to consider their offers. The venture debt thought the best bet was to back the Solaris effort as a new company. There was some interest from the big names in storage, but being a network storage insurgence means you don't fit in with the other network storage companies products. But they sure grabbed up all the engineers.

So now Coraid is back thanks to SouthSuite, how did you get the opportunity to restart this? Obtain the domain name, IP, etc.?
The new company offered us the classic Coraid software. They only wanted the Solaris bits. The Classic Coraid stuff, the stuff that ironically sold $100 million, they didn't want. But they didn't want to let the Coraid brand go. They hung onto the domain name and the trademark. I bought just the software. Later, the Solaris company folded and abandoned the trade name and domain so we were able to grab them. The company is SouthSuite, not Coraid, Inc. Coraid as a company is dead. SouthSuite owns the mark and I have built the new company with almost all new staff.

What is the plan now? Are you self financed?
I am busy working on all the stuff I had worked on during the California Coraid days but couldn't get shipped. We are funded by friends and we are profitable. We now have the time to do all the things we want to do. No money clock to IPO. We can afford to see which of our ideas are useful. And we have a lot of ideas. I like being the black swan that springs technologies out of nowhere, like NAT and VPN.

What was the reaction of the installed base? Did you sell to past customers new generation of products?
People are thrilled. They breathed a sigh of relief. They are back to looking smart for choosing Coraid. Some had to buy other products and found the experience painful and expensive. They are very glad not to have to do it again to buy more storage. We support all their existing Coraid appliances, even the appliances California Coraid stopped supporting. We will never not help a Coraid customer.

What is the current product line?
SRX 8 is shipping, VMware 6.0 and 6.5, Windows, and Linux EtherDrive HBA is also going out. VSX is going to ship later this year. We have a new family of hardware from Supermicro to run it all, the "h" series. The rest I can't talk about. There is certainly a lot in the pipeline.

Are prices still as attractive as they were?
More so. Our basic unit is the SRX RAID appliance software. We sell both the software and the hardware it runs on. We price the software about the same as RedHat, VMware, and Microsoft, and sell the hardware basically at cost. A 24 bay chassis costs $1,995. Just add software, disks, and extra NICs. It costs out much much cheaper than anything else including cloud storage. Five year cost can be as low as $0.001 per gigabyte per month.
Our basic unit was the SRX RAID appliance. We sold both the software and the hardware it runs on. We price the software about the same as RedHat, VMware, and Microsoft, and sell the hardware basically at cost. A 24-bay chassis costs
$1,995. Just add software and extra NICs. It costs out much much cheaper than anything else including cloud. Five year cost can be as low as $0.001 per gigabyte per month. We also will also ship our network storage virtualization appliance, the VSX later this year. It lets you do some pretty cool things with the SRXs. All of this works with our EtherDrive HBAs that makes all this look like just a local disk drive.
We keep making the software more efficient which allows us to use less expensive processors. The range of processor prices are about 10x. Other folks need really beefy processors to work at all. Simple code has many benefits. Being faster on less expensive hardware is one of them. Simple is just harder work and requires talent to make, that all. That's why you don't see much if it.

What about future features and technology direction? Such NVMe and NVMf?
New media is always exciting. NVMe makes a lot of sense. SSDs are good but the SAS and SATA link layer is a huge bottleneck. NVMe widens the interface using PCIe. It all works great over AoE.

And the historic Coraid channel, did you restart business with them? Did you start a brand new channel? What about OEM?
We did. A great many resellers are great at helping their local customers and helped them get and use Coraid. We have a couple of ways that makes sense for them to help their customers. We don't have any OEMs at this point.

What will be the next steps in the next few quarters?
Expect to see a lot of new things. We are always about making life simpler, cheaper, and easier. We will make the existing stuff more efficient but we won't mess it up with needless modifications. But there are a lot of other appliances and features and media to make great use of.