Panzura: 'Infrastructure is slowly getting commoditized'

We talk shop with Ranajit Nevatia, vice president at the U.S. cloud storage company.
Written by Andrew Nusca, Contributor on
Ranajit Nevatia speaking at the Structure Data 2013 conference in New York.

U.S. cloud storage startup Panzura was founded in 2008 with a mission to bring cloud-equipped local network-attached storage to large companies with operations on multiple continents.

The missing word in that sentence? Scale. With storage capacity exploding and companies leaping into new geographies, it's that rapid growth curve that makes easy file accessibility a challenge.

The company is well funded, having raised $33 million in five years from Khosla Ventures, Matrix Partners and Chevron Technology Ventures, among others.

I checked in with VP Ranajit Nevatia to find out how enterprises are responding to his company's cloudy overtures.

ZD: So, cloud storage. Are companies still scared of moving things to the cloud?

RN: Cloud storage was something a lot of people had in the back of their minds last year, but they were not doing something specific with it. This year, very large enterprises in multiple, different verticals are adopting the benefits and economics of deploying it.

Everybody claims to be a cloud vendor these days, including my laundromat, which outsources my laundry service. When we say "cloud," we mean challenging existing norms to build highly scalable, really elastic storage capacity. The 800-lb. gorilla here is Amazon with their S3 product, object storage delivered as a service. We are a front end to that.

Whether in an OpEx or CapEx model, people are now starting to really adopt it. We're seeing success in the government vertical; the [U.S.] Department of Justice has 265 sites, each managing 250 terabytes of storage, that they're now consolidating in a hybrid cloud [situation]. It's now petabytes of storage. We are enabling that. In this case, we're deploying with EMC's Atmos product.

On the Amazon front, we are seeing adoption in multiple verticals. Architecture and design—DSNR in New York, which was affected by Hurricane Sandy and were able to stay up. We are seeing success in healthcare and media. DreamWorks has multiple terabytes right now, archived, in HP's cloud service. They are storing their media assets: movies. So their sites collaborate on those assets, to people can make changes and share them.

ZD: Sounds all well and good, but which kind of companies are holding out? Why?

RN: I don't think there are any particular ones that are holding back per se. Insurance companies are the toughest or slowest to move; they still use a lot of mainframes. Pharmaceutical companies are slow to move, because they have a lot of regulations.

ZD: And for those companies that do adopt, what's their primary reason? Is there one?

RN: We are enterprise-focused, as opposed to other players that are SMB-focused. The main driver that I'm seeing is really cost reduction, headcount reduction, the move from CapEx to OpEx that grows with you on demand. It's also the case of where customers are trying to focus more on their core competency. Where does their IT department bring value to their end users?

Typically, storage is a necessary evil. You don't want to do it if you don't have to. What you are really delivering to your end users is the capability for them to be more productive with whatever application they're using. Backup and archive are usually not core competencies. So they're trying to move that to as-a-service and make it somebody else's problem.

The second trend I'm seeing is massive consolidation. Consolidation into a centralized infrastructure is, whether in a public or private context, I totally believe that, as far as infrastructure is concerned, [consolidation] is happening, and it should happen. Infrastructure itself is slowly getting commoditized. Infrastructure-as-a-service is still not a commodity. How you deliver it, what you guarantee, what your SLAs are how you manage all that? It's not yet commoditized.

You get that gigabyte of storage out of S3, and you get a similar gig from another provider that might be using similar components but is not as stable. So that's a premium. A gig from Amazon and Google: are they similar? What about a gig of storage deployed in-house? I would say there might be little differentiation when it comes to technology, but more when it comes to the service. Google makes different claims, because of their global reach, than Rackspace. Everybody has their own differentiations.

ZD: Given this, what's next for Panzura?

RN: We are just scratching the server surface here, to be honest. In terms of penetration, we have a long way ahead of us and a huge opportunity in terms of going broader within each of those verticals and going deeper. We do have three distinct use cases: many companies start with backup, and move into the NAS use case, and as they get more comfortable they start deploying across their enterprise. We continue to do that with our customers.

Customers start with 100 terabytes. They don't immediately start with a petabyte. But we're starting to get to those levels.

Everybody wants cloud storage because they see the economic benefits of it. But there are four or five key objections across industries. We call ourselves "objection-removers." All of our partners are out there trying to sell their cloud storage.

ZD: Let's talk about those objections specifically. What are companies telling you?

RN: The first concern that comes up, always, emotional or political, is security. Dealing with the government, security was the number one requirement. Three companies were invited to the bid, but two dropped out, just because they couldn't address the security problems of the government.

Our founders come from a security background. In a startup case, going after the government is unusual. Very few actually get the FIPS 140-2 validation and certification. We went through NIST, which ended up becoming a customer.

In some cases there are technological challenges or regulatory requirements. For instance, I've been in conversations where they say that their SLA to customers says that their data is not going to leave their four walls. And no matter what the case is, that has to happen. That requirement in the grand scheme of things is a really small percentage [of customers]. It's getting better and better. DreamWorks was very careful and did a lot of due diligence in securing their media assets. Security teams do get involved. Our sales cycle is started by the cloud provider, not us.

The second big objection we get is availability. Clouds go down and up. Networks go down and up. Cloud storage—object storage—is highly durable because of the way it's designed. It makes multiple copies of data. It gets geo-dispersed. You get eleven-nines of durability. But not availability. If you store a file with me, I guarantee I will get it back to you. But I can't guarantee that I can bring it back in a second or minute or hour. With Panzura, we can guarantee availability even with the networks down or cloud storage down. Because the devices sit local to the datacenter and end users, we can do a lot more intelligent things to make that data much more available to that end user.

The third big thing is latency and bandwidth of the cloud. How do you put hundreds of terabytes into the cloud from your datacenter?

The last one, which is very important, is easier to solve but not that easy, either. Cloud object-oriented storage does not speak the language of the enterprise. When DreamWorks is using video-editing software to edit media assets or the [U.S. Department of Justice] is working with a large case document, these things don't talk to the back of the cloud. You and me, we work with files, not objects. There's a big disconnect between an enterprise application and object storage. We have successfully converted objects into global files. So we deliver that object storage in a form that's usable to you.

Amazon today delivers two trillion objects. It took them six years to get to the first trillion, and six months to get to the second trillion, I think. If you look at who is consuming that, do you think it's the Fortune 500? Or is it someone else? The biggest tenants are consumer-focused, web 2.0 companies. Not the enterprise.

It's become a very proven model for deploying lots of storage in a very cost-effective way. What they're all trying to do now is push it into the enterprise. And to do that, they all have to address those four objections.

ZD: And are they?

RN: Last year, we started seeing a big move for large enterprises to start adopting this new platform. They now know how to do it.

We have two kinds of experiences now: the first is when we start with the backup admin. It's not a smooth ride for us, because he is in no way going to eliminate his own job. So we get into longer sales cycles when we start there.

But when we start with the CIO...well, the CFO is breathing down the CIO's neck right now: a dollar a gig when I can get it for 10 cents a gig? The CIO is looking to modernize his infrastructure and keep himself relevant to the business units. They look at archive as low-hanging fruit. Some of our sales cycles started at the CIO level close in six months—multi-million-dollar deals.

Editorial standards