Back in the early 1990s, with a $75,000 grant from the National Institute of Health, Gaurav Dhillon literally started Informatica in his garage. During his 12-year tenure as CEO, he built Informatica into a public company that stands as one of the leading EAI (enterprise application integration) software vendors.
In Silicon Valley, this type of success tends to breed a hunger for more. Dhillion invested in several startups, but in his area of specialization, integration, he felt there was little new under the sun. That changed in the mid-2000s when he happened to serve on a board with Marc Andreessen, who talked at length about the way people were using the cloud for their work lives. A light came on and Dhillon asked: "Gosh, how are we going to connect that? How does that work?"
[ Also see "How to integrate with the cloud" by InfoWorld's David Linthicum. | Read "Cloud computing's missing link" for more on public cloud integration provider Boomi, acquired late last year by Dell. ]
So in 2006, Dhillon invested in and helped build SnapLogic, a startup that delivers a cloud-based version of EAI. In 2009, Dhillion stepped in to become SnapLogic's CEO.
Cloud application integration may not be the sexiest topic, but for enterprises in particular, it's one of the gating factors in cloud computing adoption. Without it, when companies adopt SaaS applications, for example, they simply move information silos outside the firewall and create old problems anew. We began the interview zeroing in on exactly how SnapLogic was intended to solve the cloud integration problem.
Eric Knorr: When you helped start SnapLogic, what was your unique value proposition? Why did you think new technology would be required?
Gaurav Dhillon: We felt the solution should in itself be cloudy. It should be able to scale up to much more data -- big data, if you will. Also, the new solution should be able to scale out to hundreds of potential applications, not just "n," which is what we had in the client/server era.
We have a cloud solution that has a visual way of expressing the flow of information amongst various cloud or on-premise applications. We call them pipelines. They're really visual representations of Unix pipes, but they're designed for business users -- easy to use. If you can use Visio you can drag and drop the stuff around. That may be connecting your financial system (say, on-premise SAP) to your cloud CRM system (say, Salesforce).
The SnapLogic cloud integration platform interprets that pipeline and does the actual connections -- of orders, invoices from the financial system to customers, salespeople, commissions in the CRM system, and so on.
Knorr: How does your solution differ from other cloud integration platforms, such as Boomi?
Dhillon: Our approach is aimed at the enterprise. What does the enterprise need? The enterprise needs to connect on-premise applications -- whether they bought them like SAP or built them themselves like many of the trading systems on Wall Street or like many of the custom manufacturing systems in high tech. They've either built them or bought them, but they typically have a three-tier application feel to them, much like the era in which they were built. Connecting them to the popular cloud applications is SnapLogic's sweet spot.
The way that SnapLogic works is through hub-and-spoke architecture. Our scientific breakthrough is the invention of the standardized container for connecting any application -- be it Twitter or SAP or dBase IV files coming out of a point-of-sale terminal.
We have a way of representing that internally in SnapLogic in a neutral canonical format. We call that a Snap. We have a way to Snap, say, Salesforce into the SnapLogic hub -- and then you can Snap in NetSuite on the other end of it. So you have a connection from point A to point B. You therefore need the Snaps, the two spokes in the hub, A and B.
So the hub and spoke of connecting A to B in a neutral canonical format now lets you connect B, C, D, E, F, and G. For example, one of our customers is moving from one financial package to another financial package -- and nobody died in the making of that change. Which is amazing! Imagine that, it's a public company that's going to replace one financial system with the other. How often does that happen? It's very expensive, it's very hard, people don't do it. But if you use SnapLogic, you can simply Snap out the old financial system and Snap in the new financial system as your business grows and you need to move to a more scalable financial package.
I use the analogy of a standardized shipping container, which has revolutionized the world of transportation -- it's the same shipping container on a boat, on a truck, on a train. In that same manner Snaps connect all the way from on-premise, legacy, packaged ERP up to the cloud, including social media like Twitter and so on.
To take that logically further -- because we've invented this way of having a standardized container -- we then opened it up to third parties and built the SnapStore, where we have over 110 Snaps today, most of which are third-party. That combination of the Web architecture in the SnapLogic server, a visual representation, and hundreds of Snaps gives us a unique competitive advantage.
Knorr: What's the difference between Snaps and ordinary adapters?
Adapters really have no intelligence baked into them. They're just helping you move from one type to the other, without adding a neutral canonical format. All an adapter is doing is what the pipe adapter does -- it takes one diameter to another. It doesn't really add anything.
SnapLogic's Snaps -- for example, the Salesforce Snap -- is more like a network sniffer. It's designed from the ground up to operate across the public Internet. It has the ability to go to Salesforce, look at the customizations of the Salesforce objects, inspect the directory, inspect what's going on in there, and give you back a simple visual view of your Salesforce information without having to read the 500-page programming manual for the Salesforce API.
Knorr: Is that your target market? The technical business analyst? It sounds as if you really want to make integration more accessible.
Dhillon: I think you really put your finger on it. Most integration products are designed for programmers. Snap applications are designed to be broadly usable by nonprogrammers -- i.e., business users, who in today's age are graduates with a lot of computing skills when they join the workforce. This is not my father's businessperson, right? This is not the VP of HR from two generations ago. This is a VP of HR who went to college with a notebook computer as part of the required list. This is someone who's pretty savvy. What we're doing is giving them technology to help them be a self-helper when it comes to connecting cloud applications together -- and connecting those cloud applications to the on-premise systems that they possess.
Knorr: Do Salesforce integrations make up a good portion of what people use SnapLogic for?
Dhillon: It does indeed. We have many customers with Box.net, some with NetSuite and a variety of other SaaS applications, but certainly Salesforce is the majority.
Our installed base is by and large young public companies or large enterprises. The difference is the way in which the spokes scale out to solve the integration problem in a high-performance kind of way. The key to think about us is an enterprise-grade technology along the lines of what you might buy from an IBM or a Tibco -- potentially even some new products from my old company.
Knorr: How does a startup like yours break into the enterprise market? Do you think the time is right for this? Is it your reputation from Infomatica?
Dhillon: Having a resume does help. CIOs are very dismissive of small companies. But look, we cracked the code with Informatica, too, right? Ultimately you come back to the disruption. If you actually get to look at the architecture and you get to use the product, you realize this is a real platform for integration built in the age of the Internet. It is designed to work across relational data and semi-structured data; it is designed to work in the local area network and an Internet network across you and your providers of platform and storage or software that are being accessed as a service or the public Web. That is very important to the enterprise.
Ultimately I think the prior success does help. I do believe, though, that to win you need a disruptive architecture, a disruptive solution -- and we have that.
And I think the SnapStore really rounds it off nicely. Everybody knows the iPhone app store concept, and us being able to open up our platform in much the same way that the iPhone was opened up to really compete and dominate that category, we think sets us apart in a very interesting way.
Enterprises love it because they can build private Snaps. Enterprises have, in addition to performance requirements, a desire to extend a repeatable, reusable integration choice. And it's very straightforward to use the same API, the same SDK, that was used to build one of the Snaps in our SnapStore -- whether we did it or somebody else did it -- for an enterprise.
A pharmaceuticals company may want to build a repeatable integration for proprietary applications that manage clinical drug trials for them. And also to intersperse in that information management to assert certain numbers of regulatory compliance rules. So that information is gathered and at the same time they're complying with the regulatory body that they have to comply with in the geography in which they operate to perform those drug trials.
So the private Snap concept is the killer Snap.
Knorr: So you can create a private app store for your enterprise?
Dhillon: That's exactly right. So you get the n-squared network effect of all the Snaps in the public SnapStore -- and you can enhance that with your own. Same development environment -- Eclipse or NetBeans or whatever -- same Java or Python programming libraries to create 15, 20, 60 lines of code to write a Snap by using our prebuilt platform. But now you can choose not to publish it and reuse it inside the company.
Knorr: Are we talking mainly about data integration rather than process integration?
Dhillon: That's a very good question. We don't distinguish. I would say we're almost 80 percent process and about 20 percent data, because, look, the Internet does away with the quaint distinction of app integration versus data integration. You're basically connecting things together.
Knorr: And in the cloud, the need to connect things together seems to be growing. Lately I've been hearing more and more about a rising level of interest in the public cloud.
Dhillon: This is the year of the enterprise cloud. I would boldly proclaimed from any rooftop, if they would let me, that 2012 is the year the enterprise cloud.
Knorr: What exactly do you mean by that? Public cloud? Private? Hybrid?
Dhillon: I mean the first time enterprises use the public cloud in a big way. And therefore, by definition, it's hybrid, right? Because up until now they've been holding back.
Knorr: Why now? Why not last year or two years from now?
Dhillon: My take is that it's a wisdom-of-the-crowds type thing. It's a human psychology thing. Two years ago you have a dinner with a bunch of CIOs and they're all trying to see who had gone to the cloud first -- because none of them want to be first: "Well, what are you thinking about? Well, maybe I'll do CRM this year. Maybe we'll move that helpdesk ticketing system." But nobody was doing anything in a big way.
But now, your dinner with maybe six or eight CIOs, they don't want to be the last one to do it.
Knorr: Do you think there's an economic imperative as well?
Dhillon: I think that's a very good observation, because it is now three years since the downturn of 2008, and a lot of things were deferred. In an economic downturn, everybody freezes up. Some things got freed up so there was a little bit of a spree. Now some of the longer cycle investments are now coming -- that were delayed or people weren't really thinking about it hard.
But also I'm just seeing IT budgets being slashed across the board, because people expect everything to get faster, better, cheaper. And you know what? Everything does, except enterprise software.
It may not be too complicated. It really comes down to very basic fact that using the cloud itself to build new cloud solutions -- to build robust, mission-critical, enterprise-grade stuff -- is faster, better, cheaper. Salesforce has demonstrated it, Workday has demonstrated it, and I believe we have the potential to demonstrate that.
This article, "Interview: Gaurav Dhillon, cloud entrepreneur," originally appeared at InfoWorld.com. Read more of Eric Knorr's Modernizing IT blog, and for the latest business technology news, follow InfoWorld on Twitter.