Talend Reflects on DI, MDM, and the Open Source Advantage

Several data integration players market all-in-one DI suites -- complete with ETL, DQ, and MDM -- but Talend does so at open source price points

It's been a busy 2010 for open source data management (DM) specialist Talend, which launched its first ever master data management (MDM) offering -- Talend MDM -- in January.

Last week, Talend introduced an update -- version 4.0 -- of its DM suite, trumpeting what it says is a creditable combination of DI, data quality (DQ), and MDM in a single platform.

Although a number of other players can point to similar DM features, officials concede, Talend can do so at open source price points, then adds open source amenities (such as source code escrow) for good measure.

Not many customers actually download Talend's source code, concedes Yves de Montcheuil, vice president of marketing with Talend. Nevertheless, he maintains, the knowledge that it's out there and publicly available is, for many customers, a kind of psychological comfort.

"The truth is that very, very few people take advantage of that" ability to view or download source code, he admits. "For a lot of customers, it's important to know it's there, but when you look at the download stats in our source code repository, very few people access the source code. They can if they want, but most don't do it."

Source code escrow makes for a particularly compelling pitch in the Age of the Megavendor, according to de Montcheuil.

"I don't think too many people are aware of vendor lock-in until they run into a serious conflict with their vendor, or until they run into a commercial thing that goes sour. When you're working with an Oracle or an IBM, they're [vendors] that can afford to walk away from a $1 million customer," he points out. "When people have run into those kinds of issues with a vendor, then they are a lot more mindful of the [potential for] lock-in."

Talend's New MDM Play

Talend's MDM entry is its newest and (perhaps) most intriguing gambit. With its previous offerings -- ETL, data profiling (DP), and data quality (DQ) -- Talend found itself competing in established markets.

Take DI, for example. By the time Talend debuted its open source ETL engine -- in late 2006 -- the DI market had coalesced around a handful of megavendors (IBM Corp., Informatica Corp., Oracle Corp., and, less prominently, SAS Institute Inc.), with a sprinkling of smaller or specialty players (Ab Initio, Evolutionary Technology Inc., Pervasive Software) together with a few larger, non-DI focused players -- chiefly consisting of the former Business Objects SA (which marketed ETL, DQ, and data federation technology) and Microsoft Corp. -- rounding things out.

Although the DI market's established pecking order arguably made it more difficult for Talend to be competitive at launch, its commoditization -- which was, in part, a product of Microsoft's and Oracle's decisions to bundle ETL technology with their RDBMS products -- turned out to be the thin edge of the wedge.

Talend couldn't credibly claim parity between its Open Studio product and DI offerings from IBM, Informatica, Oracle, or SAS; on the other hand, it could claim to target the same kind of good-enough ETL segment that Microsoft had addressed with both its first-rev ETL offering (Data Transformation Services) and SQL Server 2005's SQL Server Integration Services (SSIS).

Consumers of good-enough ETL didn't (and still don't) need the bells, whistles, and price tags of the premium DI suites; many were (and are) looking to tackle basic, one-off, or predictably repeatable DI tasks. (Even as late as 2005 and 2006, "ETL" in many shops took the form of script-driven FTP transfers, with transformations facilitated via programmatic SQL and loading accomplished via least-common-denominator interfaces such as ODBC or JDBC (see http://tdwi.org/articles/2005/04/27/when-does-enterprise-etl-make-sense.aspx for more information).

The MDM market, on the other hand, isn't fully mature. It's still in the process of consolidating. It's arguably more business-focused -- and less fundamentally techie -- than either the ETL or DQ markets. Although the term "open source" was coined to appeal to (or to allay the concerns of) business buyers, the Oxford-wearing business user isn't the first user category that comes to mind when one thinks of the term "open source."

Although the MDM market's comparative immaturity amounts to an opportunity for Talend, it doesn't necessarily mean there won't be problems -- for example, the lack of a clearly-defined "good-enough" feature set. The flip side of that coin, says de Montcheuil, is that it also gives Talend an opportunity to get in at the good-enough get go, so to speak.

"You don't do disposable master data management. That's clear. But you can do tactical MDM. If you need to manage your IT assets or your application users, that's not enterprise-wide MDM that's going to consolidate all of your assets into a single behemoth repository. It's a project that can start small and grow. You start with a half-dozen applications, build the hub, build the integration between the hub and other applications, and go from there," he explains.

Nor does de Montcheuil view MDM's more business focus as a drawback, either. Talend MDM isn't the product of a loose confederation of open source coders, he points out; it's an enterprise software company with dozens of customers and several successful rounds of venture capital financing behind it.

More to the point, he maintains, it's able to recruit and retain business talent. "Master data management, probably like data quality and a little bit like data integration, is typically implemented by IT to serve the business. While it's true that it can be challenging [for this reason], we have hired some very knowledgeable people in the MDM space, stolen from the competition," he asserts, citing recent defections from both Initiate and IBM.