John R. "Beau" Vrolyk, President and CEO, 3ware
"Ethernet is everywhere, and that ubiquity is what will eventually force Fibre Channel on to the scrap heap."
October 11, 2001
![Network Computing logo Network Computing logo](https://eu-images.contentstack.com/v3/assets/bltde8121fc52c5c8f3/blt3f3d0318f746b1c2/65a530e4187606040a1d8b8c/placeholder.png?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
Most people in the industry agree that the jury is still out on whether IP will even get close to replacing Fibre Channel as the de facto standard for storage networks.
Not Beau Vrolyk. Hes so sold on the idea that in 1995 he came out of retirement to be the CEO of 3ware Inc., an IP storage startup based in Mountain View, Calif. 3ware makes an IP storage box called the Palisade that contains an Ethernet switch and the capacity to hold multiple terabytes of storage.
He was sailing his boat around the tranquil islands of New Zealand with his family at the time. "The Internet was just beginning to take off, and I couldn't stand missing it," he says. "I'd had enough of floating around in the South Pacific and became very grumpy." Eventually his wife and kids agreed to head back to the U.S. (for a bit of peace and quiet, by the sound of it).
According to Vrolyk, (who doesn't appear to be insane), Fibre Channel is going the way Token Ring did: under. IBM spent millions of dollars marketing Token Ring and still only got 10 to 15 percent market penetration, he says. (Although FC is the dominant techonlogy in storage networks today, it has only penetrated about 30 percent of the potential market.) And the reason for Token Ring's demise? “Ethernet, Ethernet, and more Ethernet,” Vrolyk says. “It’s everywhere, and that ubiquity is what will eventually force Fibre Channel on to the same scrap heap.”
Should he be wrong (gasp!) and IP storage doesn’t get off the ground, he still has the boat and will go back to sailing it. But he's confident that won’t happen: “The Fibre Channel folk waffle on about TCP/IP not being robust enough for storage applications. It’s totally irrelevant."Click on the following links to find out why:
Byte and Switch: Can we clear up the “Beau” thing? What’s up with that? "John" not good enough for you?Vrolyk: I am either the fifth or sixth John Vrolyk in a row. My 13 year-old is also a "John Vrolyk." As a result, if you yelled "Hey John!" in our house, you might get anywhere from 1 to n (depending on how many of my ancestors are alive and on-site). As a result, almost everyone has a nickname. Mine is "Beau," which is a result of having spent a few of my early years in Tennessee.
[ed.note: Better than Bubba, right?]
Byte and Switch: Okay, Beau, so what's all the fuss about iSCSI [the soon to be ratified SCSI-over-IP standard]?
Vrolyk: iSCSI will do the same for storage networking that HTML did for Web browsing. Like HTML, iSCSI isn’t elegant or gracious in the way that it works – it’s simply going to be everywhere.
Every single time a debate over protocols comes up in networking, the market goes for the lowest common denominator. It doesn’t need elegance, just ease and efficiency.Byte and Switch: But the major corporations that have deployed SANs today have built them using Fibre Channel. In the current market climate they are even less likely to replace this gear with new, relatively untried equipment.
Vrolyk: Right, it isn't going to be an overnight job, but the market conditions are in fact an incentive for companies to gradually roll out IP storage technologies, as they're so much cheaper than Fibre Channel. It's the major corporations that have the most to save that are checking out this technology. And don't forget: The SAN market is barely out of diapers yet.
Byte and Switch: What about InfiniBand? That’s beginning to cause a few ripples in the SAN world as a way to connect storage in large data-center environments.
Vrolyk: Yawn! – or just YAN, really. [ed.note: Beau says this stands for Yet Another Network.] What are you going to do with the 300 million or more computers already connected via Ethernet? It’s back to the Token Ring problem: It was a better technology, but people were already using Ethernet; they understood it and saw no additional benefits that made them want to change.
I think the macro trends in our industry are frequently obscured by a number of quick pops (like flash bulbs) of different technologies that flame and die. InfiniBand is one of them.Byte and Switch: 3ware's equipment provides block-based storage access only. But as the convergence of NAS and SAN happens, you'll eventually also provide file serving as well, I take it?
Vrolyk: No, we won’t actually, as this is a myth. I don't think that NAS and SAN will merge – they have different uses. For example, databases run only on block interfaces, while lots of things like your home directory are best built on top of file systems. For years, block and file interfaces have existed side by side in server operating systems. I don't think that will change.
Rather, the question is: Should the same box be both a file and a block server? If you add the extra computing to run a file system to the base cost of every block server, it will simply be too expensive. Watch and see.
Block servers will be much less expensive in a few years than any file server. This trend is obscured by the fact that the only block servers available today, other than 3ware's, are Fibre Channel based and, as a result, are ridiculously expensive. Once 3ware and others provide high-volume block storage on Ethernet, it will be much more economical than file storage.
Byte and Switch: So why are so many of the startups – and EMC Corp. (NYSE: EMC), come to that – talking about converging the two?Vrolyk: They are reaching for a way to look better and to differentiate themselves. We have picked one thing [block serving] and we do it extremely well.
Byte and Switch: Doesn’t virtualization [which gives administrators a single view of a collection of diverse storage devices] imply NAS and SAN convergence?
Vrolyk: Nope, another myth. Virtualization presents a view of the storage space, not the actual data itself. So it's not true convergence.
Byte and Switch: You were one of the turnaround team brought in to sort out Silicon Graphics. Did you achieve that?
Vrolyk: No one ever succeeded in turning around SGI. I survived three CEOs in three years there as head of engineering and then product marketing. It wasn’t easy.Byte and Switch: How has what you did at SGI helped you in your position at 3ware – or, more diplomatically, what does a graphics company know about storage?
Vrolyk: SGI built some of the best and largest Fibre Channel SANs for the entertainment industry and the U.S. government.
Byte and Switch: Ah, okay... Perhaps we should be talking to them then?
Vrolyk: No, don’t bother. SGI has other things on its plate right now, like how to claw its way back from financial ruin.
Byte and Switch: We noticed from your bio that you did a philosophy degree. Has that affected your thinking at all at 3ware?Vrolyk: I was trained in symbolic logic, which is really useful when writing software and solving problems. I can't help you with the Meaning of Life. That's the other sort of philosophy student.
Byte and Switch: Yeah, right, the really dull kind. Here's a quick test of your powers of philosophy: Why did the chicken cross the road?
Vrolyk:... Because the Margaritas were over there?
Byte and Switch: Correct! Very good.
So... Where does the time go?Vrolyk: Well, time is truly –
Byte and Switch: Ah... no. That’s the end of the interview, actually, but thank you.
— Jo Maitland, Senior Editor, Byte and Switch http://www.byteandswitch.com
You May Also Like
Maximizing cloud potential: Building and operating an effective Cloud Center of Excellence (CCoE)
September 10, 2024Radical Automation of ITSM
September 19, 2024Unleash the power of the browser to secure any device in minutes
September 24, 2024