News Stay informed about the latest enterprise technology news and product updates.

DataDirect CEO: Storage needs are unpredictable, demand scalability

Clustered storage vendor DataDirect Networks has gained a foothold in the rich media and Web 2.0 markets that larger storage vendors like EMC and NetApp are trying to move into.

DataDirect Networks might be the biggest storage company that most storage administrators have never heard of. DataDirect CEO Alex Bouzari says his clustered storage systems company has more than 160 PB of its clustered storage installed at customer sites, which include AOL, NASA, Time Warner Cable, and Ford. The company's OEM partners include IBM, Dell and Sony.

Even though DataDirect generated $100 million in revenue last year and has been profitable for four years, DataDirect remains privately held – with plans to go public early next year. The company was primarily a niche high-performance computing (HPC) player before riding the wave of digital content in media and entertainment, gas exploration and Internet companies over the past few years. Now DataDirect is poised for competition from the large storage companies looking to attack the Web 2.0 market. recently spoke to Bouzari about DataDirect's plans.

More on storage vendors
DataDirect launches 1 PB-plus SAN system  

HP previews Web 2.0 storage system   

EMC ships Hulk, says multiprotocol storage to dominate
NetApp fleshes out plan for OnTap GX clustering convergence  
SearchStorage: Who's buying your storage?

Alex Bouzari: We started in the scientific verticals – our customers had hundreds of terabytes and petabytes and needed scalability. We also have a lot of customers in rich media and high performance computing and the last several years we've seen unstructured data making its way into the domain of lots of large organizations.

We see a paradigm shift to the world becoming more digital. We developed a platform to handle the needs of unstructured data – images and video and sound – and where the Internet is involved, people need a large repository for huge amounts of content.

Storage needs have also become more unpredictable. Xbox Live is stored on DataDirect, and at any point in time they don't know how many users will be saving game levels of Halo 3. That unpredictability requires scalability tfor storage that can grow by an order of magnitude.

SearchStorage: What is DataDirect's main value to storage shops?

Bouzari: We're the only architecture that can write data as fast as it can read. We're designed for unstructured data and real-time environments. Most storage out there is designed for static environments and transaction data. We're seeing a fundamental change in storage and data and content. It's no longer about random bits of data in databases handling things like stock market trades. It's much more about hybrid environments that are content-rich – videos, photos, rich media files. And the sizes of environments have become large and dynamic – storage needs are unpredictable.

If you were to take our appliance and plug it into traditional storage environments with smaller I/O, that aretransaction-based and random in nature, we would suck. We would be horrible. Conversely, take a system designed for those type of environments and put it into unstructured real time scalable environments. They would suck. Our competitors are trying to retrofit their systems for unstructured systems.

SearchStorage: Who are your main competitors?

Bouzari: Until three or four years ago, there wasn't much competition in HPC and rich media. We were the only game in town. As we broaden into new markets, we see EMC. We see NetApp where there's a NAS element. In rich media, to a much lesser extent, we see a little bit of Isilon but typically our low-end products overlap with their high-end. The other competitor is LSI, mainly because LSI is sold by some of our OEM reseller partners, IBM, SGI and Cray.

SearchStorage: What about XIV Ltd., the company that IBM recently acquired?

Bouzari: IBM happens to be one of our OEM customers. In terms of where they're positioning XIV, it seems to be in different areas and in different markets than where they're positioning us. XIV seems to be designed for offline, not real time, and storage that's static in nature.

SearchStorage: What's your relationship with IBM?

Bouzari: We signed an OEM agreement with IBM last August. They're selling our storage appliance under the IBM brand [as the IBM System Storage DCS9550 Storage System for high-performance computing]. They are most probably going to be our No. 1 customer for this year. We have significant wins with IBM, including oil producer Saudi Aramco.

SearchStorage: You use the term "extreme storage" to describe your systems. How do you define extreme storage?

Bouzari: For us, extreme in performance means the ability to power the fastest computers in the world. We're also extreme in density – square footage is limited and expensive. We can pack more capacity and more bandwidth per square foot than anybody else. Last year we did a few installations north of 10 PB.

SearchStorage: You talk a lot about speeds and feeds of your systems. What about data protection and management software? Do you support features such as snapshots and replication?

Bouzari: Going forward as we broaden our portfolio and get it to mainstream, we're working on tools along the line of replication and snapshotting. Today we offer that in partnership with other companies, but we're developing them natively inside the appliance. But there are quite a few tools that have been implemented to manage the system. Our systems can read and write at the same speed, even in degraded mode. That's because our roots are in streaming video and rich media.

We have developed a suite of applications that are focused on management. We have an operating system that resides inside the application, called DirectOS. DirectOS manages storage – the storage is basically pooled and we can provide access privileges to different customers. We do firmware upgrades across the whole population of disk, and have predictive management tools. We also have a sleep mode capability where customers can selectively put drives to sleep.

SearchStorage: Do you do support other features that are becoming common in storage systems, like thin provisioning?

Bouzari: We're looking at thin provisioning, but that's not really a value-add in the markets we serve. Our customers are more concerned with the ability to quickly store content, have it readily accessible, and have the ability to scale. Thin provisioning is useful if you're dealing with a legacy storage environment.

SearchStorage: You do support MAID. How is your disk spindown unique?

Bouzari: The main difference in our MAID implementation is it's a dynamic MAID, you can spin down disk and LUNs selectively, based on whatever parameters you might choose. It's not a static or rigid MAID. It can be modified if our customer needs to change it. We have customers using us for primary online storage as well as archiving. They can pick drives in a pool and spin those down and keep others spun up.

SearchStorage: You say your drives are self-healing. How does that work?

Bouzari: As data comes into our appliance, we separate the data from commands, and we have the ability to check parity on the fly. SATA drives are inexpensive, but they also exhibit a nasty effect of silent data corruption -- they're corrupting data but you don't know it. We can catch this data correction on the fly, correct it on the fly, deliver the right data to the user and fix the problem in the background. We can do partial rebuild on a drive. You have terabyte drives now, if you have a small piece that needs to be rebuilt, why rebuild the whole thing? We can rebuild just the piece of it that needs to be rebuilt.

SearchStorage: With $100 million in revenue and years of profitability, you're in a better position that most of the storage companies that have gone public the last few years. How come you haven't filed for an IPO yet?

Bouzari: We're planning on an IPO in nine to 12 months. It's a matter of market conditions. Being profitable and growing nicely, there's really no impetus to get it done sooner and try to launch into a market that has a lot of volatility. If the market were to stabilize sooner than nine to 12 months from now, then we can go sooner. Why take the company out now and take a 40% haircut on what the valuation should be?

We grew from $25 million in revenue in 2003 to $100 million last year. We've seen an accelerated need for our type of storage. We broke even in 2004 and have been running profitably since then. We had limited venture capitalist backing, and we bought out our investors a couple of years ago. We didn't say, "Let's go out and raise $100 million or $200 million and hope at one point we have enough top-line [revenue] to break even."

SearchStorage: DataDirect has been referred to as a well-kept secret. Why don't more people know you?

Bouzari: Up until a year and a half ago, we felt it made sense for us to fly under the radar mainly because we were selling into segments of the market where there wasn't much competition. Why make a lot of noise when we're doing great? When we started going into broader markets we said, "OK, now's probably a good time to start building up the company." At the same time, tier 1 companies approached us about selling our products. With people starting to notice us, we said we might as well come out of the closet and let everybody know who we are and what we do.

Dig Deeper on Unstructured data storage

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.