I talked with the SnapLogic team last week, in connection with their SnapReduce Hadoop-oriented offering. This gave me an opportunity to catch up on what SnapLogic is up to overall. SnapLogic is a data integration/ETL (Extract/Transform/Load) company with a good pedigree: Informatica founder Gaurav Dillon invested in and now runs SnapLogic, and VC Ben Horowitz is involved. SnapLogic company basics include:
- SnapLogic has raised about $18 million from Gaurav Dillon and Andreessen Horowitz.
- SnapLogic has almost 60 people.
- SnapLogic has around 150 customers.
- Based in San Mateo, SnapLogic has an office in the UK and is growing its European business.
- SnapLogic has both SaaS (Software as a Service) and on-premise availability, but either way you pay on a subscription basis.
- Typical SnapLogic deal size is under $20K/year. Accordingly, SnapLogic sells over the telephone.
- SnapReduce is in beta with about a dozen customers, and slated for release by year-end.
SnapLogic’s core/hub product is called SnapCenter. In addition, for any particular kind of data one might want to connect, there are “snaps” which connect to — i.e. snap into — SnapCenter.
SnapLogic’s market position(ing) sounds like Cast Iron’s, by which I mean:
- As a practical matter, clients usually first buy SnapCenter to connect on-premise and SaaS applications.
- SnapCenter supports cloud-to-cloud* and on-premise-to-on-premise integration as well.
- SnapCenter itself runs either on-premise on in the cloud. (Larger customers at the moment tend to prefer on-premise deployment.)
- SnapLogic suggests its products are simpler than many ETL alternatives.
Not atypically, SnapLogic believes that SnapCenter is higher-end than Cast Iron (which is now an IBM company), and that SnapCenter’s real top competitor is in-house/hand-coded integration.
*When discussing data integration, “SaaS” and “in the cloud” are close to synonymous.
What SnapLogic said about its use cases seemed to boil down to:
- The base SnapLogic use case is when a client wants to push all data from one application to another application.
- The source and target applications can be any combination of on-premise, SaaS (e.g. salesforce.com), and so on.
- ETL purchases often start when somebody purchases a database (by which I presume SnapLogic also meant data feed).
- SnapLogic sees three main kinds of use case:
- One-time batch (a big move of historical data).
- Classic repeating batch.
- “Trigger-based” (I don’t think SnapLogic was using the term “trigger” just in its technical DBMS sense).
- SnapLogic sees a big future in providing a scalable integration layer for Twitter, RSS feeds, and so on, especially straight into websites, but I get the impression there are only a few pioneering users of such capabilities right now.
The main technical sizzle in the SnapLogic story is the SnapStore, with lets you download free snaps and buy unfree ones.* SnapLogic says there are 100 or so snaps in the SnapStore more, with a couple more being added weekly. That claim started making sense to me when SnapLogic said most snaps are offered by system integrators (as byproducts of specific integration contracts?) or software vendors (to connect to their own offerings?).
*I was expecting snap pricing to be subscription-based also, but when I went to the SnapStore this didn’t seem to be the case.
At least, I think that’s the main sizzle. I’ll confess to not having come away with much understanding of other nuances of SnapLogic technology. In particular, I don’t know what the core data interchange format is that allows all the “simplification” and “normalization” needed for this approach to be possible. So in particular I didn’t drill down far enough to uncover any limitations (functionality or performance) in that aspect of the architecture, and the same goes for SnapCenter’s RESTfulness. That’s all probably my fault; SnapLogic did put a bunch of good people on the phone, and we did at least lay the groundwork for future understanding.