I get the impression Informatica is still Salesforce-only, e.g. from this price list.

Pervasive DataCloud is currently vendor-specific too. In Pervasive’s case, the fixed point is QuickBooks Online. DataCloud, the pure cloud offering. is newish, with an undisclosed hosting partner. The most common integration is with, you guessed it, Salesforce.com, but Microsoft’s CRM is in the mix as well. Pricing is $1-2K/year.

The most comprehensive integration-as-a-service story I’ve heard may be the one Cast Iron Systems is rolling out. Cast Iron is hosting with OpSource any integration you can get in the Cast Iron appliance. To emphasize this, pricing is identical to that of the rental option for the appliance ($1K/month in the simplest two-endpoint cases), and customers are encouraged to switch between appliance and cloud usage as they see fit. (That said, I think the whole thing is way too new for such a switch ever to have happened yet; the official rollout is scheduled for October 20.) Cast Iron supports a fairly broad range of applications, SaaS and on-premise alike. (Cast Iron is particularly proud of what sounds like a beyond-Barney hug from Oracle’s CRM On Demand business.) Cast Iron claims less than a handful of direct sales of this new cloud offering. However, Cast Iron also claims 23 partners, combined from among several areas:

• SaaS vendor OEMing for ongoing data integration in the usual way
• SaaS vendor OEMing for one-time data migration
• Implementation VAR using the service

You may have noticed that everything I’ve cited above is for operational apps being connected with each other, almost always including CRM. What I haven’t heard is integration vendors getting much involved with analytics-in-the-cloud offerings, whether from data mart outsourcers or vendors with cloud DBMS offerings. Not coincidentally, I don’t think many offerings in either category have large customer counts. (Also — Kognitio, which along with Vertica is one of the two data warehouse DBMS vendors most emphasizing cloud offerings, happens to have a data migration subsidiary of its own.)

1. An appliance delivery format.
2. Lots of heuristics for automatic mapping and quick set-up. E.g., Cast Iron claims that 70% of a typical SAP-Salesforce.com connection can be done straight out of the box.
3. The absence of data cleaning/transformation features that might complicate things.

Cast Iron still believes in all that.

Even so, its messaging has changed a bit. Cast Iron now bills itself, in the first sentence of its press release boilerplate, as “the fastest growing SaaS integration appliance vendor.” And when I talked with marketing chief Simon Peel today, the only use cases we discussed were connections between SaaS and on-premises apps. Simon’s contention is that Cast Iron is much better suited for such uses than, say, industry leader Informatica, which he positions as being big, cumbersome, feature-laden and, above all, focused on batch processing. Indeed, Simon thinks Cast Iron has many more SAP-to-SaaS connection customers than Informatica. (I’m guessing that set includes an outright majority of a total customer base he says is in the 2-500 range, a figure that includes both end users and some OEMs. Other comapny metrics include 200%ish annual growth in both customers and revenue, and a little over 100 employees.)

Informatica and Cast Iron both seem to be on big pushes to gain partnerships with application software vendors. But their focus is different. Informatica is after — well, they’re after everybody, at least above a certain size threshold, and that threshold is going down. Cast Iron, however, is specifically focused on SaaS providers. To that end, they’ve proudly partnered with Opsource, and hence at least implicitly with however many hundreds of small app vendors are using Opsource to port their software to SaaS.

So what’s to keep Informatica or IBM from squashing Cast Iron? Simon fondly believes that he has a whole complexity/cost/business-model disruption narrative going. (My words, not his.) I think that’s somewhat overstated. Integration is fundamentally an add-on rather than platform technology; any vendor that has temporarily forgotten how to sell on-the-cheap ala carte tactical solutions could easily re-learn how.

So any sustainable advantage for Cast Iron probably lies in competitive barriers to emulating its simplicity. Cast Iron, in essence, is all wizards and (at least ideally) no code. Can a more fully-featured competitor put a wizard layer on a code engine? Sure. But Cast Iron seems ahead in actually building the wizards, at least in its target niche.

Among the companies they cited, you can find topical categories here for IBM (and Cognos), Informatica, Microsoft, Netezza, Oracle, SAP/Business Objects (both), SAS, and Teradata; QlikTech; Cast Iron, Coral8, DATAllegro, HP, ParAccel, and StreamBase; and Software AG. On Text Technologies you’ll find categories for some of the same vendors, plus Attensity, Clarabridge, and Google. There also are categories for some of these vendors on the Monash Report.

There’s a more substantive disagreement as to whether the features deleted from BOBJ’s product due to the injunction are actually important in the marketplace. I’m looking into that subject, and hope to post about it in the near future.

There has not yet been any appeal. There is no new decision. And BOBJ had no choice in accepting or rejecting the remittitur – what the BOBJ rep referred to as the “judge’s new proposal.” Only Informatica, the prevailing party in this case — has the option to accept or reject the remittitur.

To save a round of “he said, she said,” let me head off at the pass another point BOBJ may proffer. While the judge found that the $25 million dollar award was appropriate and supported by the evidence, the remittitur reduced the award to $12M because case law had changed as a result of AT&T v. Microsoft. The ruling in that Supreme Court case came in after the INFA v. BOBJ verdict and held that the jury could not include damages for sales of copies made outside the United States. Specifically, the case resulted in a legal change that holds that plaintiffs relying on US patent rights can only go after US damages and would not be entitled any international damages. The decision had a binding effect on our case and the Judge, therefore, gave Informatica the choice to retry the issue of damages again next year (and assume all of the costs associated with this – and be limited to only evidence of US damages) or take a reduced damages number of $12.2 million.

We want to be clear that: 1. Informatica could have asked for a new trial to determine a new damages amount; 2. Though Informatica may have obtained even more in a new trial, $12M for BODI’s sales in the United States represents a very substantial royalty of approximately $20,000 per infringing sale.

In summary, the reduced amount had nothing to do with the merits of the case or any “appeal”’ by BODI – it only had to do with unfortunate timing of the AT&T case and that decision and the fact that given some administrative issues a final order had not yet been issued when the AT&T case came down. There is no new decision – the jury’s verdict stands (other than the amount) – the patents are valid, BODI willfully infringed those patents and Informatica was awarded an injunction against BODI.

The final order will be entered in the next few weeks and then they will appeal the decision – which given the facts in this case – will be a tough road ahead for BODI/SAP.

Keep getting great research about data warehousing, analytics, data integration and related technologies. Get a FREE subscription by RSS or email!

Technorati Tags: Informatica, Business Objects

• I recently talked with Pervasive Software about their data integration line. A large part of Pervasive’s new business is Salesforce.com integration, including at some big-name software vendors as customer/partner switch-hitters.
• I just rechecked my notes from my January talk with Cast Iron Systems. A large part of Cast Iron’s new business is also integration with Salesforce.com, Netsuite, and other SaaS vendors.
• Informatica keeps putting out press releases about Salesforce.com integration, most recently by offering replication in SaaS form itself.

But of course this makes sense. Without good data integration, SaaS applications would be pretty useless, at least at large and medium-sized enterprises.

First of all, they’re extending their connectivity tools to work well across enterprises. We didn’t talk about the detailed technical implications of this, but I’d guess there was a fair amount of security-related work, and a fair amount of additional effort to handle actual remote networking (session management, traffic shaping, etc.). Second, they are putting software onto the Salesforce.com platform itself, with all the multitenancy/ zero-footprint requirements that entails. This seems to be mainly in the data cleaning/quality and replication areas, at least at first.

As simple as this sounds, I think it’s of considerable market importance. Without good data connectivity, frankly, I’ve never seen why SaaS made much sense for large enterprises. I think a lot of CIOs would tend to agree with me. This kind of technology will make it a lot easier for IT shops to embrace SaaS.

Key product advantages they claim include:

Integrated active metadata directory. This is apparently based on Informatica’s own code, rather than a licensed product. It is used a lot at development time, not so much at run time, with the obvious application to Sarbanes-Oxley-style compliance being exploited only by relatively few customers. I must confess to chuckling that Sohaib Abbasi has wound up running a company that boasts of its active metadata directory. Such a directory lay at the heart of Sedona, an Oracle dev tool fiasco of legendary proportions,.

Note: Notwithstanding my crack about Sedona — by almost any standard, Sohaib is one of the good guys. (I was going to put “surprisingly” in that sentence, but actually the same is true of most other ~20-year Oracle veterans too.)

Performance. Informatica pooh-poohs anything XSLT-based as being inherently slow, due to its document-orientation or something. I.e., such products are row-at-a-time rather than set-at-a-time. However, it didn’t occur to me to probe for a convincing argument as to why set-at-a-time is inherently faster. Whoops. We have a follow-up item.

Data cleaning/quality versatility. Informatica acquired the Similarity product some months ago, which they assert is more modern than some competitors, and hence better suited to handle data beyond names/addresses. A key example would be product hierarchies/ taxonomies. I suggested they explore whether this could be leveraged for enterprises’ text technology architectures, specifically in the area of ontology management.

Data transformation complexity. Here too I was sloppy about figuring out what the exact claim was. Performance? Ease of development/administration of the transformations themselves? General goodness of the scheduling, etc. tools? Probably it’s some of each, especially the first two. In any case, “complex data transformation” commonly gets intoned when Informatica tries to position its products as being higher-end than, say, Business Objects’ EIM offering.

Lots of connectors. Informatica has been around for ages. Accordingly, it connects to almost anything.

SaaS/Outsourcing. This really does sound like cool stuff.

Informatica talks about wrapping/enabling data services, in a way that would warm a DBMS2 implementer’s heart. Where they don’t play, however, is on the process side — they position themselves away from EAI and ESBs, they don’t have a composite app dev tool, and so on.

Frankly, I think they still have a marketing problem. Data connectivity middleware is boring, and they don’t have a high-concept story to jazz it up.* I think they could, and have pointed them at some of the original DBMS2 materials for clues about how to articulate it. But they don’t have one yet. It also doesn’t help that Informatica, for many years, was one of the most visibly for-sale companies in the whole software industry, although they vigorously deny that that posture continues under Sohaib’s leadership.

*However, they do get beaucoup Brownie points for not inflicting on me the story “We offer an abstraction layer that liberates you from being locked into various different DBMS brands (and locks you into our software instead).” In various forms, that pitch has been annoying me for a quarter of a century. Indeed, they derided it as just selling a “nicer jail.”

First, it’s a marketing construction — a blend between between ETL (Extract, Transform, Load) and EAI (Enterprise Application Integration). It’s a legitimate category; all those things are getting smushed together as near-real-time apps become more prominent. Still, it’s also an attempt to grab marketing turf.

Second, it’s commonly associated with a marketing overreach — the claim that an EII “platform” or “suite” will do everything a DBMS does (almost), but fully and heterogeneously distributed as well. Yeah, right.

Third, two of the sharpest proponents have been acquired by behemoths that tend to obscure their acquirees marketing pitches — Ascential by IBM and SeeBeyond by Sun.

Fourth, some of the best grand integrated EII suites (at least the ones that started as ETL, which is the side I’m more familiar with) aren’t complete yet. So vendors didn’t want to be too clear for fear of freezing current sales. I’m referring here mainly to Ascential and Informatica. They told analysts of their grand plans, but they haven’t been so eager to openly publicize the full details.

Fifth, the area is getting integrated with development tools for composite applications. Good examples there are SeeBeyond and Intersystems’ Cache’.

Sixth, no EII vendors’ plans fully work unless they have full relational and XML integration, and nobody really has been doing a great job on that, typically being strong in one area or the other.

Obviously, this is an area I have to research actively; EII is the neuromuscular system that holds DBMS2 together. But all the research in the world won’t change the fact that as of now it’s the weak spot in the story. There’s lots of great database management technology, and lots of excellent reasons to use a variety of kinds of that technology in your enterprise. But the tools to knit the resulting heterogeneous databases together are still sadly deficient.