Higher Ground

Ascential seeks upper echelon of the ETL market with Torrent acquisition

Like the rest of the broader integration market, the extract, transform, and load (ETL) market has been moving toward realtime performance. Ascential Software Corp.'s purchase of privately held Torrent Systems Inc. for $46 million may signify an opening up of the highest echelon ETL market — one heretofore dominated by Ab Initio Software Corp. — a rarefied realm in which large batches of data are integrated extremely frequently into a data warehouse for nearly instantaneous analysis.

Frequent ETL batches, as frequent as one every second, are particularly useful to CRM systems. For large enterprises with heavy data volumes, such as telecom companies, frequent batches don't necessarily mean small batches. Gartner Inc. predicts that increasing e-business transactions will require very large corporations to manage hundreds of terabytes of data by 2005. Torrent's Orchestrate is a framework for high-volume, parallel data integration and analysis (particularly CRM analysis) that Ascential began integrating with its DataStage XE product in March 2001.

But while the newly enhanced Ascential seems poised to jump into the high-margin game that Ab Initio has been playing as solitaire, doubts and reservations about its success may hold it back. For one thing, analysts question the terms of the deal. It's not clear why Torrent, which had just completed another round of venture capital funding and seemed to be doing well, would want to cash out. And the outright cash transaction, with no stock component, would seem to give the Torrent crew little incentive to stick around and erase the seams between Orchestrate and DataStage XE.

Philip Russom, an independent analyst and contributing editor to Intelligent Enterprise, counters the last concern by saying, "Integrating products here means sharing metadata capabilities, tightening communications between products, and pulling design and administrative tools into a single console. This is light work compared to folding source code together."

For background on the evolution of the integration market toward realtime performance, see Philip Russom's Intelligent Enterprise feature article "Beneath the Waterline," May 7, 2001. For background on the role of massively parallel systems in scalable data integration, see Russom's "Mass Movement," May 24, 2001.

— Jeanette Burriesci

In this Issue: