While director of IT for Motorola, Chet Phillips led a team that Intelligent Enterprise honored with the 2003 RealWare Award for Best Strategic Business Performance Management Application. Now an independent BI consultant and self-described activist, Chet took a break to offer some opinions on the current buzz in BI and data warehousing.
Q: Is data warehousing becoming obsolete because of the push for real-time intelligence?
A: Based on my conversations with numerous business leaders, I don't believe that the real-time surge will take off for probably another two years... and years in the IT world are like dog years. I'm searching for that first real-time killer app that I need for my business, but my business users are telling me they don't need it at this point.
There's a lot of rumbling going on about the EII [enterprise information integration] space. It would be nice if we didn't have to take data and put it in a new spot like a data warehouse, but right now that does add value because of the way that data warehouses inherently try to normalize the data and put it into a spot where you really can compare apples with apples and oranges with oranges. Until we get enterprise metadata more squared away, we won't be able to do that dynamically.
Q: How important is the service-oriented architecture based on Web services?
A: It's another area where everybody's all dressed up but has nowhere to go. I do believe that 2004 will probably be the year of Web services, but I don't believe that it'll end up being what everybody thinks it's going to be where you can create a service and then put XML behind it. Frankly, it goes back to the metadata stuff.
Q: Grid and utility computing: Is there any real benefit to BI applications in them?
A: Yeah, actually. There is still this inherent overengineering that goes on with enterprise applications and data warehousing. You tend to make them bulletproof, fault tolerant, disaster recovery worthy... So you put more money into them than you actually get out, from the hardware perspective, because nobody wants to screw up the enterprise app. So I could see grid computing easing some of that, the cost side of the return on investment for BI.
On top of that, there are some very interesting technologies going on in the ETL space that will allow dozens of different threads to go in parallel in the grid environment. And believe me, we've run into situations where we're pulling so much data that the hardware becomes the constraint. Some of my peers tell me about how they've had to use all of the performance tuning they could to pull in their terabyte a night.
Q: How about consolidation of BI vendors? Do you think that's helping people like you: directors of IT and CIOs?
A: BI customers are watching the industry to see where there's going to be overlap in a specific vendor relationship. They're watching Brio and Hyperion to see if they step on each other, and what investment is going to be put into each one of those products when they both work for the same people.
So there may be a temporary balking at buying new tools because people are trying to understand whether product lines are going to be dropped or not. Business Objects and Crystal Decisions is another example of where people in the industry are still trying to figure out whether that even makes any sense at all.
A lot of IT professionals who are in the position to buy these tools need to have some concrete evidence as to what the newly created or consolidated companies are going to do with those product lines. If they in any way, shape, or form go and support a specific initiative and then all of a sudden they [the vendor] end-of-life it... Most IT professionals still have whip marks on their backs from some of the ERP and CRM stuff that went in. So we've kind of run out of favors and we're looking for that sure bet. The company that comes to the table with the most concrete view of how this stuff is going to work and interoperate with other products, is the one that'll get the lion's share of the business.
I think the winning vendors will be able to be open on their standards; they'll have to be a lot more agnostic. And another good trend I'm seeing is that some companies now are decoupling. They've put together a whole solution and tried to sell it as that, and now they're decoupling that solution so that people can go and put together their BI solution a la carte.
I thought that we had gotten past the proprietary thing a while back, but it keeps rearing its ugly head. But most of the IT directors are coming up with architectural governance standards that will disqualify these proprietary vendors when they try to come into the enterprise. I don't want to trivialize it, but it's a very fashionable thing right now in large enterprises to come up with your chief architect and your architectural standards. It's probably way overdue.