Sleepycat to extend paw into Asia

Company president speaks

Sleepycat Software is looking seriously at setting up operations in Asia and has ambitious hopes for its XML (Extensible Markup Language) and Java embedded open-source databases. The company introduced version 2.0 of its Berkeley DB Java Edition (JE) Monday.

IDG News Service chatted with the head feline at the 30-person privately held firm, Sleepycat President and Chief Executive Officer Mike Olson, about the company's genesis and the statistic Sleepycat loves to throw around -- the estimated 200 million deployments of its Berkeley DB software. The firm's customers include, America Online (AOL), Cisco Systems, EMC, Google, Hewlett-Packard and Motorola. Sleepycat pioneered a dual licensing model for its products, a no-cost open-source license allowing redistribution if the application using the database is open source, and a commercial license for the redistribution of proprietary applications.

Olson was in at the start of what would later become Sleepycat, being one of the original authors of Berkeley DB, while studying at the University of California at Berkeley. He then worked at object relational database company Illustra and later its purchaser Informix Software. Olson joined Sleepycat in 1998 as vice president of sales and marketing, becoming company president and CEO in 2001. The company's headquarters are in Lincoln, Massachusetts, and it has offices in Emeryville, California, and Woking, U.K.

What follows is an edited transcript of the interview with Olson.

IDGNS: Where does that 200 million figure for Berkeley DB deployments come from?

Olson: We have a very large installed base. It's north of 200 million today. The number may sound outrageous, but every single copy of Linux, Solaris and Mac server includes Berkeley DB, also products from vendors like Nokia and Ericsson include our database. We got someone to measure our installed base some years ago, but we haven't redone it. The number getting bigger doesn't help us from a business perspective.

IDGNS: So, what's the Sleepycat story? The company was founded in 1996.

Olson: You have to go further back than that. In 1991, Keith Bostic was a staff member and Margo Seltzer and I were grad students at the University of California at Berkeley working on database technology for Michael Stonebraker. We were working on BSD [Berkeley software distribution, a derivative of AT&T Bell Laboratories' Unix]. Keith needed to replace all the AT&T proprietary software with the BSD Unix and he was working on a component DBM [database management] in the Unix library. Keith knew Margo and I particularly well and asked the two of us to write a replacement library. [Bostic is Sleepycat's vice president of product development, while fellow cofounder Seltzer is the company's chief technology officer.]

We wrote the original version of Berkeley DB in 1991. Then I totally forgot about it. I joined Illustra with Stonebraker [llustra's founder]. Keith and Margo continued to maintain and evolve Berkeley DB throughout the early 1990s. It was pretty popular and freely available and lots of projects chose to use it. By the mid-90s, it was pretty big and really grew by word of mouth and there was increasing demand for an enterprise upgrade. Users needed more reliability and recovery as opposed to the academic version of the software.

In 1996, Netscape was flush with cash from its IPO (initial public offering) and went around buying up groups with interesting software. They pulled in people from the University of Michigan working on LDAP. They used Berkeley DB and liked it. The problem was it was single user and crashed. They suggested Netscape should fund the work, and, after some negotiation, the deal was struck. You can think of it as a seed financing round for Sleepycat.

IDGNS: And, I have to ask, where did the name Sleepycat come from?

Olson: The story I have from Keith and Margo -- they're both cat people -- is that they'd gone so far as to finish the software and sign the commercial license with Netscape. They needed to incorporate the company as part of the deal. All the paperwork was filled in but the name. They'd been working with their cats sleeping on their monitors, so they chose Sleepycat, thinking they'd fix it up later. And here we are nine years later. The company's well enough known that it'd only be an embarrassment to change it now. The name's a little bit more frivolous than I'd like, but if it's OK for there to be a Yahoo or Google, Sleepycat's not so far outside the pale.

IDGNS: Sleepycat opened an office in the U.K. in November of last year. Do you have plans to set up elsewhere in the world, given the particular interest in open-source software in Asia, Eastern Europe and South America?

Olson: We are seeing increased adoption of Berkeley DB, especially in Eastern Europe and Asia, China in particular. I'd expect that to continue to be true. Our next significant move will to be into Asia. We've narrowed our shortlist to China, Japan, or Australia/New Zealand, which is probably an easier place to set up a business. If we go into China, we'll find a local business partner and open up a channel relationship.

The U.K. team is able to service Eastern Europe, we already have deals in Poland and Moscow. We only have a few South American licenses so far. We'll go into South America some time next year.

IDGNS: At the start of 2004, you publicly announced that Sleepycat's revenue growth in 2003 was 60 percent higher than in 2002 when the company grew by 35 percent. You didn't release such statistics for 2004, how did you do?

Olson: In 2004, we had just about 36 percent revenue growth over the previous year. Our history shows a good strong revenue growth of between 30 and 40 percent, growing every single year in total revenue, profitability, and the number of customers and repeat customers. We've never taken outside funding. So far, we haven't decided if we need to take outside funding as we move into other geographies.

We've no current plans to take Sleepycat to an IPO. We'll continue to build our business and broaden our product line. Our long-term strategy is less about going IPO and more about some strategic M&A [mergers and acquisitions]. We're moving particularly aggressively into the Java space. People used to build very high performance systems using C. Now it's the case that the hardware and JVM (Java Virtual Machine) are good enough that people are building large, complex systems entirely in Java.

IDGNS: How does your revenue split out between your three products -- Berkeley DB, Berkeley DB XML and Berkeley DB JE?

Olson: First let me mention that about 70 percent of our revenue comes from the licensing of our open-source databases for proprietary applications. It's paying all of the mortgages around here. We do sell services and maintenance as well.

The breakdown of revenue across the three products? Obviously, Berkeley DB is 14 years old, or nine years old, if you count from when the company incorporated, so it provides the dominant share of our revenue. Our XML and Java databases are relatively new. We backed them out quickly to get some direction from customers. In 2004, a minuscule fraction of our revenue came from XML and Java, but I'm convinced these two products are going to grow in dollar terms and in the long term will be major contributors [to the bottom line]. The way I look at it, we've ploughed the field, planted the seeds, and now we're starting to reap the harvest on these two products.

Copyright © 2005 IDG Communications, Inc.

How to choose a low-code development platform