In-Depth

Q&A: ETL Dead and Done With?

Ascential Software says it has moved from an ETL provider to a data integration company. Will 2005 be the year of metadata?

• By Stephen Swoyer
• 12/17/2003

We spoke last week with Bob Zurek, vice president of advanced technologies and product management for Ascential Software Corp., about the changing landscape of enterprise data integration. Because the focus of enterprise data integration has shifted away from ETL and toward metadata management, Zurek says, ETL isn’t quite the data integration sine qua non that it once was. It’s still important, Zurek concedes, but it’s now one process among many—others include data quality and data profiling processes—that’s needed to front a complete enterprise data integration offering.

You established yourself as an ETL provider, but you’ve made a couple of not exactly ETL-related purchases over the years [Ascential has acquired data quality, data profiling and parallel processing assets]. Last year, you unveiled what you called an Enterprise Integration Suite, of which ETL was but one component. So how do you position yourselves now?

Our focus is as a data integration company. That’s all we do. We’re very, very focused on solving this data integration problem for large- and medium-sized businesses globally. We have a very strong passion about this focus, and you won’t see us moving into portals, we’re not building a super stack, we’re not building a messaging structure. We’re building the richest infrastructure out there that handles transactional data, operational data, analytic data, and we think that Web services and service-oriented approaches are a common theme that’s front and center in today’s buyers. So that’s what we’ve done with the Enterprise Integration Suite and our Real Time Integration Services.

When I talk to you [Ascential], you’re always careful to stress that you’re not just an ETL vendor. Does this mean that you feel ETL isn’t the essential technology that it once was?

Although ETL will be a very prominent process out there, more and more people are using the product capabilities in the [Enterprise Integration] Suite to standardize on it as a data integration service. That’s why we’re transforming the nomenclature away from ETL and toward enterprise data integration. We’re delivering this message: We think that ETL is basically dead and over, and it’s all about enterprise data integration using an end-to-end approach that is highly scalable, reliable, and fits within the fabric of Web services and other standards.

But most shops aren’t yet moving to Web services, are they? Aren’t most people still using ETL to address most of their data integration needs?

The transition is actually happening very rapidly. And it’s not just Web services, but other processes, like metadata management and data quality. Customers are taking a very keen interest in metadata and data quality initiatives, and they’re not seeing it as just purely ETL.

But they’re all still using ETL, right?

Yes, obviously, but they’re using it with other processes. Some people have to take data that’s coming out of EDI systems or have to be HIPAA compliant, and that’s not just extracting data and loading it into a warehouse. Things like data quality [and] data profiling are natural extensions to ETL and are all part of it, but we think that the next pillar going forward is going to be metadata management. We think that 2005 will be the big year of metadata, when finally people get what it’s all about.

Okay, I’ll bite. What is metadata all about?

The regulatory and compliance [legislation] like Sarbanes-Oxley is going to require a metadata solution to answer questions like where did this data come from, where did you get this number, things like that. The CEOs and the CFOs have a legal and fiduciary responsibility to comply with Sarbanes-Oxley, and they’re going to be asking these questions now. Metadata management is going to be essential for this purpose.

I want to go back to what you said about ETL being “dead and over.” Is this because vendors such as Microsoft and Oracle are in the game now, too, and have commoditized ETL by incorporating it into their database platforms?

A couple [of] things with relation to the database vendors. First of all, we’re a gold-certified partner with Microsoft. Unlike our competitors, who are seeing Microsoft as a competitor, we see Microsoft as a complementary partner--

Right. But have the efforts of Microsoft and Oracle to build ETL capabilities into their databases kind of forced your hand, so to speak, such that you and some of your competitors have had to kind of reposition yourselves as data integration specialists?

Well, just because Microsoft and Oracle are doing this doesn’t mean that the need for ETL [tools from third party vendors] goes away. We believe that the pure ETL functionality of the database engine in Microsoft will grow in its capabilities, sure. But we operate in very large heterogeneous environments and run natively on the mainframe platform, where we’ve been a leader. Plus, we’re running across Unix, Linux machines, and those are areas that Microsoft cannot play, and they’ll look to Ascential to help facilitate the movement and transformation of that data.

Is it commoditization? Sure. But it’s not the reason I say ETL is dead. ETL is no longer synonymous with data integration—now it’s just one process [among others]. By the way, we can actually use DTS, if a customer decides to use it for ETL, which kind of illustrates this point.

What do you mean?

We can leverage DTS as a Web service. If someone exposes DTS as a service, we can easily interface with DTS, execute at the file system, take that data, and do some enrichment on it. But our vision of data integration is exposing these processes as services in a service-oriented architecture to provide what we call “frictionless connectivity” to them.

That’s what the Real Time Integration Services that you announced back in June is designed to do?

Yes. With Real Time Integration Services and our support for Web services, we’ve been able to deliver data integration as a service to these organizations. In the past, they may have asked for support for SOAP and XML and WSDL, and we couldn’t deliver that, but today’s platform does deliver that. We started shipping our Real Time Integration Services in October and today we’re in production and preproduction with large-scale enterprises.

What exactly does it do?

It’s something that allows customers to quickly expose any data integration process as a service. You can take an existing process and expose that existing data integration process in less than 10 clicks of a mouse. It can be at any level of complexity or a very simple process, it’s the same standard approach no matter how complex or simple the process is.

Quickly, what’s in store for Ascential in 2004?

Several things. First, major investments in metadata management infrastructure. Our whole integration suite is metadata-driven through the use of an infrastructure called MetaStage, and we’re taking further steps in expanding significantly that capability. We’ve already been in conversations with some of the largest enterprise software companies who are compelled by our vision and our strategy for metadata management.

Second, we will absolutely be changing the user experience in this [data integration] environment. We’ve been on a very lengthy mission to have one of the most innovative approaches to enhancing the user experience, so we’ve already been working for two years with this on our customers, and we’re working on removing a lot of the barriers that users run into when trying to connect inside and outside their enterprise. Sometimes, they have to deploy gateways, sometimes they see four or five diff ways of speaking to Siebel or SAP, and they wonder, what’s the best approach, should I use ODBC, should I use the Web services interface, should I use their object model? So 2004 is an area where we’re focusing on the topic of frictionless connectivity.