Q&A: Ingres CTO sees move toward open-source databases

Dave Dargo also says Ingres’ latest software is more enterprise-ready than MySQL

Dave Dargo is chief technical officer at open-source database vendor, Ingres Corp. Like many other Ingres executives, Dargo is a veteran of Oracle Corp., having spent 15 years there, including a spell helping to manage Oracle’s Linux and open-source strategy. Until the late 1980s, Redwood City, Calif.-based Ingres was Oracle’s chief rival in the enterprise arena. As a result, Dargo said in a recent interview, Ingres’ latest 2006 version is more enterprise-ready than open-source rivals such as MySQL -- and far cheaper in terms of support than Oracle or IBM’s DB2.

What was the significance of Oracle’s recent purchase of open-source software maker, Sleepycat Software Inc.? I wouldn’t call Oracle’s buy of SleepyCat a move into the open-source world. Oracle has been making lots of different acquisitions, both open-source and closed-source companies, in order to shore up its customer base. All of the open-source companies Oracle has bought own 100% of the intellectual property, in order to broaden the number of people required to pay Oracle support. What I think they’re going to tell their customers is: ‘If you want any Oracle support, you’re going to need Oracle support on all of your products.’

I think it’s a heavy-handed strategy to customers. And the most heavy-handed part is that many of these customers had chosen these products in order to avoid doing business with Oracle. And now they’re getting sucked into that Oracle machine. And that’s what is making people the most nervous.

The impact on Ingres is beneficial. We give comfort that there can be competition on the support side. That’s our real market opportunity.

The big database vendors would say that they have co-opted the dual-licensing model of open-source vendors like yourself or MySQL by releasing free Express editions. They’re starting to tout some of the customers they’ve won back as a result. Do you think open-source’s momentum has slowed as a result? I really don’t see that at all. Customers started experimenting with open-source databases because they were successful with Linux and Apache Web server. And they recognized that their license fees weren’t driving new product development. They desperately wanted the open-source business model. But the open-source databases are immature. They haven’t gone through a 30-year maturation process like Oracle and DB2.

In the long run, you may see some temporary wins [by closed-source vendors with free editions] for those users who are making pure price decisions. But I think you’re going to see more people driving toward [an] open-source business model. The real value of open-source is making it more efficient for new database features to be delivered to the market. For the leading-edge customers who still need new features, there are ways to get such features much quicker than paying license fees to closed-source vendors. When you want a change in a closed-source product, you have to lobby the vendor like you would lobby Congress.

But people like Marty Beard, an executive at commercial database vendor Sybase Inc., assert that most new product development, even in open-source databases, comes from the vendor, not via community contributions. Therefore, he says, commercial vendors are more likely to be responsive and innovative because they employ many more engineers than open-source firms like yourself or MySQL. Do you agree? Some Wall Street companies told us that they have lobbied Oracle for 10 years to add certain new features. So it’s not just who has the capacity, but who has the willingness to do it. And while Oracle or IBM may appear to have more developers, the fact is the number of people actually touching the kernel is very small, usually about a dozen or so. We have a similar number of people here.

The general open-source story line is that it works by appealing to younger, lower-ranking developers in an enterprise, who install the free product at an increasingly higher level until the CIO suddenly realizes its widespread use and decides to adopt it for more mission-critical applications. MySQL and PostGres seem to have that cool factor. Ingres, being an older product that has had a long enterprise history, seems to lack it. Is it important to build that cool factor with developers again? I think that’s kind of na¿ve. It’s not the way the development organizations or CIOs I’ve worked with make decisions on what they will deploy. Developers make their decision based on what is the best tool for a specific task. Standards bodies within organizations then make their decisions on what is going to be the easiest to deploy, support and maintain.

MySQL’s popularity is at its level because it is the easiest tool to deploy Web-based applications, and it came along at a time when Web applications exploded in popularity. It wasn’t the cool factor, it was because MySQL was easy to use. But it’s a far cry from the edge to going into the core of the enterprise.

When you talk to an enterprise DBA, they don’t care about the coolness or how fast you can develop a Web app with it, they care about stability, about whether they’re going to have to come back in at 2 a.m. to recover a corrupted file. The kinds of things Ingres can demonstrate are what other open-source vendors have not even dreamed of putting in their products.

Since the spinoff from CA four months ago, what new customers has Ingres won? The deals we have closed – and we have closed many deals – have all been expansions within existing customers. I don’t know that we’ve closed any brand new customers. We’ve concentrated on communicating with our existing customers, making sure they’re satisfied. We believe that new customers will come as a natural course of how we go about the market. We’ve had many people contact us, but we’re not at a point where we can announce deals.

Last time we talked, you didn’t divulge many details about the product development. Four months later, have you formulated a product development roadmap? On Feb. 7, we released Ingres 2006, which has a new licensing, packaging and pricing model. We will also have a product coming out in the first half of 2007. We’re doing things that are subtle changes to major features we already have. Other open-source database vendors are still trying to write basic partitioning and cluster management. We’re adding things like bit-mapped indexing and temporal types of partitioning.

In the area of XML data storage, we’ll let IBM and Oracle battle it out. Ingres still stores XML data as BLOBs. I don’t want to bend the company around application-specific issues that will pull us away from the big picture.

Ingres is the only C2-level secure database out there. We can encrypt data in movement today. Encryption of data at rest is usually at the operating system level. We’re seeing if we can put it into the open-source product.

With our community edition of Ingres, we expect to have a new release each month. Then when we roll out as a single enterprise-class release in a year, people know it’s been through a heavier testing environment. We’re also working with partners to create an Ingres-type software appliance that makes it very easy to provide support for both OS and database. We’re also talking to them about how we can share the development work.

We haven’t decided yet whether to open-source the OpenROAD, Enterprise Access and EDBC products.

How is the search for a permanent CEO to replace Terry Garnett coming along? Last time, you mentioned interest in a high-profile executive on the level of former PeopleSoft CEO Craig Conway. Terry is still looking. He wants to make sure he finds the right CEO. As Terry watches the management team gel and grow, some of his search criteria may have changed.

Copyright © 2006 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon