As a small private company competing with EMC, Symantec, Dell and Hewlett-Packard (HP) in a market that some see is in decline, ExaGrid certainly faces a big challenge in the target deduplication appliance field.
But ExaGrid CEO Bill Andrews says his company is growing fast, with more than 1,800 customers, because it has the only scale-out architecture in the industry and it takes a different approach with dedupe than its larger competitors.
How has the disk backup market changed since you started at ExaGrid in 2005?
The cloud was going to paint your house, pay all your bills off and fund your retirement account. It was going
to do everything.
Bill Andrews: I've seen a lot of change in eight-and-a-half years as CEO. There was a period in 2010 to 2011 where a lot changed for us and confused the market, and that seems to be clearing now. First, Data Domain got out ahead of the market and really made the market. When EMC acquired Data Domain, they put a huge salesforce and customer base behind it. Then other companies also came out with backup hardware products -- Symantec, HP and Dell all came out with products, and we were faced with four big companies as competitors.
On top of that, there was a lot of noise about the cloud for a while. The cloud was going to solve every problem; it was going to fix everything you wanted to fix, and on top of that, it was going to paint your house, pay all your bills off and fund your retirement account. It was going to do everything.
But a couple of things have now changed for us. The first thing is the cloud has virtually disappeared from our market. Customers realize they don't have enough bandwidth to move their backup data in and out of the cloud. So the cloud has relegated itself to SMBs, the 30- to 200-person companies that have 30 GB, 50 GB or 100 GB.
Our smaller customers are 5 TB and above, and most of our customers have 10 TB and above -- up to 70 TB or 80 TB. They just have far too much data to be moving it in and out of the cloud on a daily basis.
Large companies use the cloud for other IT processes. Why can't cloud backups work for them?
Andrews: Once everybody did the math and looked at the cost of it, they found it expensive compared to doing it yourself. When people look at the cloud, they freeze for a while; they don't do anything, they do research. And things slow down. For a couple of years, a lot of people investigated the cloud. They found [that] the cloud is not right when you get to terabytes of data.
So what are their on-premises options?
Andrews: As the cloud has dissipated, it's becoming clearer to the customers that there are two architectures for appliances. You can scale up with front-end controllers with disk shelves and inline deduplication. Or you can scale out, which is what we're doing with full appliances going into a grid architecture and a landing zone, where we keep the most recent backups. The message is starting to break through.
The HPs, Dells and EMCs all have an identical architecture -- they all have a front-end controller, and they just add disk shelves as they grow. They don't bring any more processing or bandwidth or memory when the data grows. So as the data grows, the backup window gets longer and longer because they're just adding capacity -- they're not adding any other compute resources or bandwidth.
We're only ones who do it our way. If we had the exact same inline approach as the other guys, we'd be toast. Why would customer buy from us if they could get the same exact architecture from them?
We have to get the customer to say, 'I want scale-up, I want performance along with capacity, and I want the most recent backups.' If we get customers to say those things, we can beat EMC.
How is your architecture different?
Andrews: Our story is scale-out -- we add processor, memory, bandwidth and disk. We bring more compute with capacity, and that concept is really starting to play in Peoria. Plus, we're the only vendor who has a landing zone for recent backups in their complete form. So if you ask us for a local restore or a fast off-site tape copy on the weekend or a virtualized instant recovery, the data is sitting there ready to go. The other players all have dehydrated or deduplicated data, and it takes them longer to put the data back together.
For customers, here's the choice they have -- if the backup app has dedupe, then they can decide if they want to go straight to disk and do the dedupe on the backup app, or they can go with an appliance. If they decide they want to go with an appliance, then it's the scale-up approach or scale-out approach. Backup software with dedupe to disk only works for low retention -- up to four weeks -- because the deduplication rates on backup software are a pittance compared to deduplication rates on high-powered target-side appliances. So you quickly burn through disk behind a backup application, and there is a crossover point where, as the retention gets to a certain point, it's actually less expensive to build a dedicated appliance than to go straight to disk.
Do customers see the difference in your approach, or do they consider all disk targets the same?
Andrews: It's starting to pick up for us. We brought on 79 new customers in the quarter [of 2013], our second-best quarter ever.
You say the cloud has disappeared from your market, but aren't you working with a partner to provide cloud backups?
Andrews: We've created a relationship for a hybrid cloud for our smaller customers who want fast recovery. We have a relationship with ATScloud. Customers put an ExaGrid appliance on-site, and if they don't have a second building or don't want to rent space at [colocation] site, they can replicate to a second ExaGrid at one of ATScloud's locations. They charge by gigabyte per month protected. But it's really a hybrid cloud, not an Amazon type of public cloud. It's pure DR: The backup app is still on-site with ExaGrid, and the cloud is for pure DR.
ExaGrid started out by selling to smaller customers. Have you moved into the enterprise?
Andrews: We're selling larger grids. We've been moving the top end of our system up. Back in 2006, a 5 TB customer was a big customer for us. Now, we have nine models that mix and match in a grid. The largest can take in a 21 TB full backup, and we can put 10 of those in single grid. We continue to push that up every year, and you will continue to see us do that. But we're not competing with the biggest Data Domain boxes yet.
Symantec and others are having success with integrated appliances that bundle software with the disk targets. Have you explored any partnerships with backup software vendors to come up with an integrated appliance?
Andrews: We do a lot of work with Veeam. With VMware, Veeam [instant-recovery features] allows users to continue to work off backups like primary storage. You can call for the file, so if the file was backed up last night, you're getting a version of the file as it was no later than last night from a recovery point objective perspective.
Because we have the landing zone, we're sitting with a most recent virtual machine backup in its complete form. So Veeam can move the physical VM image off the landing zone. Our competitors have to wait until you rehydrate again.
We'll be doing other things with Veeam. We're also doing lot of work with CA ARCserve D2D, and we're working on other backup relationships.
What's ExaGrid's biggest challenge?
Andrews: We'll never be able to match EMC, Dell, HP and Symantec for size. But it's a big market, so we have to get into increasingly more fights. We're increasing our inside sales machine, adding marketing and adding sales territories. The more we do that, the more fights we get into. That allows us to keep growing.
Do you have to become a public company to compete long-term with the big guys?
Andrews: Our goal is to take it to [$80 million in annual revenue], and that's when we can take it public. Once you're public, that allows you to have more presence and more reach. Will we ever be No. 1? Probably not.
How far away are you from $80 million annual revenue?
Andrews: We're still two or three years away. We're on the path; we just have to keep methodically growing.