Q&A: Why Acopia didn’t get a Christmas card from Network Appliance or EMC

CEO says file virtualization company is winning over big customers, but some storage vendors see it as "Antichrist."

File virtualization company Acopia is no longer a start-up. The Lowell, Mass., outfit, launched five years ago and backed by $85 million in venture funding, has been shipping its boxes since 2004 and winning over customers such as Bear Stearns and Merrill Lynch. The company also is taking dead aim at storage stalwarts such as EMC and Network Appliance. CEO and President Chris Lynch met recently with Network World Executive News Editor Bob Brown. Here's an edited transcript of the interview:

What are your ARX switches actually replacing?

By design, nothing changes. We plug into the Layer 2 distribution fabric and the NAS devices still hang off the Cisco network and we plug in, almost like a one-armed device. We're a proxy for all file access, so the transactions between the client and backend look the same. We're arbitrating resources and adding a layer of intelligence -- through metadata -- to enable policies to better manage that backend. When you create this virtualization layer you're severing the physical connections between the clients and backend. From an operational perspective, you no longer need to touch the backend when you take the clients down. On the capex side, we can do a number of things.

In a world where everything is physically mapped you will find at Fortune 1000 companies an aggregate utilization rate of about 30% of their storage. That's because every application or business unit has its own storage and there's no way to spread those resources over who needs them. Once you create this virtualization layer we're able to pool that storage and increase the utilization rate, recapture storage for customers and reprovision on-demand.

The other impact we have is to allow customers to efficiently and dynamically tier the storage. We found at these large companies that not only is the storage underutilized, but also it's one big tier of the expensive stuff. It's not that they don’t know the other stuff exists, but because the provisioning in the physical world is manual and disruptive, it's more trouble than it's worth to move it.

I can't imagine storage companies would be thrilled about customers using their arrays more efficiently…

IBM is thrilled about it because they are a systems company and they just OEM NetApp storage. That's tactical for them. They're a data management company, so they love what we do because they have all these policy assets they can now enforce through the fabric of the network leveraging our APIs. But you're right, I did not get a Christmas card this year from [Network Appliance President] Tom Mendoza or [CEO] Dan Warmenhoven, or [EMC CEO and President] Joe Tucci for that matter.

The big storage companies have been investing in and developing all sorts of management and virtualization technologies. They really consider you guys a threat?

Network Appliance is most threatened because all they do is sell spindles. EMC is threatened less, but they're threatened because they view it as we're taking this high ground that controls sort of what they do and at some level commoditizes it. More  important is our strategic positioning in managing it.

The EMC NAS group views the world the same way that Network Appliance does. They view Acopia as the Antichrist because for them we're creating a heterogeneous abstraction. They're not used to that in the storage world. Cisco may have a dominant position in networking but there are hundreds of companies that exist in that ecosystem.

In the storage world it's "OK, you have this flavor or this and you're not allowed to have both or change with any frequency." The cost of switching is high because of the proprietary nature of the tools. There are tools that are reasonable for managing and provisioning but they're all out-of-band, meaning they're manual and disruptive and homogeneous. The way I position it is that we are technically very complementary because we make the customers' storage more manageable, more cost effective. There's going to be backlash against vendors touting their 60% gross margins on disk drives. It's like buying milk in a blizzard for $10 a gallon. You need it but you're ticked that they're doing it to you.

Are any of these other vendors sniffing around?

They're sniffing around, but the strategy at this point I think is, "Hey look, these guys are a bad thing for us in the near term so we should make our best effort to take them out and make them go away." But that never works because the customer votes. We have 100 Fortune 1000 customers. I've never had that in 20 years of doing start-ups. I'm always twisting arms of uncles and cousins and small innovative companies that will take a flyer on a start-up. There has already been some consolidation with EMC buying Rainfinity and Brocade buying NuView, which are appliance-based software companies that do elements of what we do.

How about Cisco? Is it eyeing this market?

They're staying out of our way until we build a big enough market. That's probably the right thing to do. Leave it to the experts and then they can decide to buy their way in. I don’t think they have the expertise at this layer. The easy part for them would be to buy a company like Acopia [Lynch previously worked for ArrowPoint, a content switch company bought by Cisco and that was started by Cheng Wu, also Acopia's founder], so they could have the technology, but I don’t think their sales people or channel would be able to sell it today.

Of course the market will mature and it will be easier over time, which is why I think them being in the [wide-area file] space makes much more sense for them now. [WAF] has to do with protocol and link optimization, and it's much simpler for the channel and Cisco to swallow.

Can you give me a sense of the installations at these big-name customer sites?

We've got in excess of $3 million worth deployed at Merrill Lynch where we're managing well over a billion files and are deployed in every data center around the world. At Bear Stearns we front-end a series of trading applications that they've got over 12,000 traders running through every day. We're in the data centers throughout New York and New Jersey for this specific application.

Goldman Sachs has us in front of their NAS farms for general-purpose file directory management and the global name space, this concept of a file-area network [Acopia is among a cadre of companies rallying around this concept within a special interest group under the Storage Networking Industry Association].

At Yahoo we front-end its Overture software, which I think is 70 cents out of every dollar they make. We allow the performance of that application to keep up with the demand, which has grown 40% in the last 12 months. They were looking at an entire rewrite of the application because they couldn’t address enough storage.

It's been a while since Acopia rolled out any new products. What's ahead?

It's a little premature. We do have a one-two punch, we have a significant lead and we protected our intellectual property around it in terms of the concept of doing in-band virtualization in the network.

There are people doing [information lifecycle management], and [hierarchical storage management] out of band either with agents or sitting on end-system platforms. What you will see from us soon is a major announcement with respect to the next generation extension of the software intelligence.

Think about this layer of intelligence we've built through an object-level switch. We can ferry lots of policies, and they can emanate from lots of internal and external assets, and we can federate them and inject them into the fabric of the network. We focused on file management to get in the door, but you can imagine if we can do files, doing block is an order of magnitude simpler.

If you think about the other things that come into data management envision a switch like ours' ability to do things like data logging, antivirus and providing these services in a file-area network that spans through a global namespace. Imagine being able to create a thin layer of metadata leveraging an index engine from Fast or Google so we can do things like a new level of compliance enforcement. Think about having the ability on the fly to make a determination based on a user ID based on the application you're accessing, that this data or person's data should be encrypted.

I'm a little surprised Acopia hasn't played up security more in the past. Why not?

Partially because we have to earn the trust of customers to get this fabric built. But if you think about it, that layer is a great place to deal with security and compliance enforcement because we see every file coming through. We're actually working on a service that will be deployed by Iron Mountain sometime by spring where they're going to provide compliance enforcement services to customers and use our little box as a CPE device. We're also going to provide disaster recovery replication services.

Anything new coming on the hardware side?

You'll see our technology having a new level of portability. Our technology will be ubiquitous. It will be running on an Intel platform in addition to on big fabric switches.

Learn more about this topic

Seven products for your next- generation network

02/20/06

Network World Video: Acopia

12/12/05

Acopia ups funding

11/03/04

Editors' Picks
Join the discussion
Be the first to comment on this article. Our Commenting Policies