Data Integrators Refocus

Trend: Infrastructure Based on Data Federation and Data Warehousing

Data integration, historically the domain of extract, transform, load (ETL) tools, is going through a transformation of its own these days. The changes are evident in recent moves by Informatica and Ascential Software Corp., two leading ETL vendors.

In August, Informatica announced it would no longer sell packaged analytics software. "The company is repositioning as a provider of the integration infrastructure you use to build applications," says Ted Friedman, analyst with Gartner.

Part of that infrastructure is a new product Informatica introduced in August called SuperGlue. "SuperGlue acts as a tracking and auditing trail for all metadata," says Sanjay Poonen, senior vice president of worldwide marketing at Informatica.

Also in August, Ascential purchased Mercator, another integration firm. "Mercator extends what we are doing," says Jim Welch, vice president of engineering at Ascential. "In particular, Mercator has an excellent real-time data transformation and routing engine."

Both companies partner with IBM, a strong proponent of what has come to be known as enterprise information integration (EII). "EII is a form of virtual data integration," says Doug Laney, analyst with the Meta Group. "Traditional ETL tools actually consolidate data from disparate sources into one physical data warehouse. EII tools create a unified view of your data, sometimes referred to as data federation."

For this reason Laney says EII is best suited for low-volume applications where performance is not critical. "EII has to spin a query out to all your various data sources, so there will be performance issues. It is not a substitute for data warehousing."

Even so, the two aren't necessarily mutually exclusive. Nelson Mattos, IBM director of information integration, thinks the future lies in using EII with ETL. "Data federation is not a replacement for data warehousing," says Mattos, "but the combination of the two can be very powerful. You can still use ETL to consolidate your data in a warehouse, but then data federation allows you to insulate your applications from details concerning data location."

And the key to taking this hybrid approach, according to analyst Janelle Hill of the Meta Group, is metadata, the contextual data that turns data into information. "It is not just about data integration anymore," says Hill. "It is about information integration, data, and context. IT executives need to understand this."

— Mark Leon

Mark Leon [mrleon@usfca.edu] has been reporting on business and technology for the last seven years.

In this Issue: