Zenoss: Bringing open source to enterprise management

Startup born from the failure of IBM Tivoli views the community as its foundation for success

If you curse your "tier-one" IT management solution as too cumbersome, too complex, and too expensive, you're not alone. In arecent Gartner study that declares the "Big Four" -- BMC, CA, HP, and IBM -- increasingly vulnerable to SaaS (software as a service) and open source alternatives, surveyed users gave their vendors mostly C's and D's, and a good number of incompletes.

For Bill Karpovich, co-founder and CEO of Zenoss, dissatisfaction with the leading management frameworks is a business model. In fact, Karpovich notes, Zenoss' open source network and system monitoring solution was born from the failure of IBM Tivoli at USi, an application service provider (and now a subsidiary of AT&T) where he and Zenoss inventor and CTO Erik Dahl used to work.

"At USi we had spent millions on IBM Tivoli, and like many people we were very unhappy that it didn't work," Karpovich recalls. "Erik was the architect that was brought in to help us figure out our management platform. Erik's recommendation was to throw out Tivoli and build our own."

And so USi built its own management infrastructure, which included both commercial off-the-shelf products and open source software, bringing all of the pieces together into a framework called USi Oasis. The experience led Dahl to realize there was a big opportunity to build a next-generation network and systems management product on the open source stack. He left USi to do just that.

The difference between Dahl's "next-generation" approach and the traditional tools, Karpovich says, is the focus on relationships among network devices, servers, and applications rather than on individual systems.

"Model-based management has been Erik's philosophy from day one, and today that is called a CMDB basically," Karpovich says. "So that was at the core of the product from the start, building that model in an object way that is extensible and supports complex relationships."

According to Karpovich, the Big Four, built back in the client/server era, simply didn't have the right data model to account for these relationships, or to provide the different views into the IT environment that enterprises need today. And their way of doing business is outdated too, he adds.

"They're living off an old version of the enterprise software model that doesn't exist today, which is selling software for a million dollars and having it take a year to deploy," Karpovich says. "We're just beyond that."

As Zenoss Core approaches its 2.0 release in June, the open source project boasts more than 13,000 registered downloads from SourceForge.Net. Among the additions in the new release are discovery of configuration policy violations and a plug-in framework for application monitoring packs called ZenPacks.

ZenPacks for Microsoft Exchange and Microsoft SQL Server will be immediately available to enterprise subscribers from Zenoss, with more ZenPacks to follow from the company and the community in coming months.

The enterprise edition also features a WYSIWYG environment for running synthetic transactions against Web applications, providing a way to monitor a Web site from the user's point of view.

"You can correlate the end-user experience with all of the element level views," Karpovich explains. "So I can say, Wow, I broke my 8-second rule on adding an item to the shopping cart, and I wonder what was happening on my network at the time, I wonder what was happening with memory and disk? Were there any processes that were taking up too much CPU?"

Zenoss' business model is based on selling professional services, support, and value-added functionality to enterprise subscribers. But Karpovich believes the company's success depends on leveraging the community to improve the product, build out the library of ZenPacks, and most of all, help Zenoss keep up with rapidly changing needs.

"That's where the Big Four struggled in the past, right?" asks Karpovich, suggesting one last lesson from history. "A technology trend occurs, and a year later you get a management tool for it. It's just too slow. The world moves too quickly.

"IT management is the infinite problem, so it's perfect for open source," Karpovich concludes. "One company simply can't generate enough ideas or enough R&D to stay on top of managing all of IT."

Join the discussion
Be the first to comment on this article. Our Commenting Policies