Open source systems management software [LWN.net]

September 29, 2006

This article was contributed by Glyn Moody

A previous LWN feature examined the rise of the open source enterprise stack - a modular collection of applications that together provide the entire spectrum of enterprise computing functions. One component of that stack is systems management. This area encompasses areas such as provisioning and patching of servers; configuration and management of applications running on those servers; and monitoring all elements of the computing system - hardware, software, networks and their security.

Systems management is dominated by the "Big Four": BMC's Performance Manager, CA's Unicenter, HP's OpenView and IBM's Tivoli. Like many proprietary systems, these are monolithic in design, and attempt to provide every kind of systems management features within a single, highly-complex program.

Free software is by its very nature modular, so open source systems management programs tend to be focused on particular tasks. This has led to a richness of the free software tools addressing this area, often with multiple solutions for a given problem. The downside is a confusing array of possibilities, a wide range of rival approaches and some unnecessary duplication of effort.

In an attempt to bring some harmony to this coding cacophony, the Open Management Consortium (OMC) was founded in May 2006 with the following objectives:

Create awareness of open source management tools in the market
Provide education and resources to help end users make informed decisions regarding open source
Establish conventions and standards that enable integration and interoperability
Enable collaboration and coordination on common development projects
Promote collaborative open source systems management solutions

The founding members of the consortium are Ayamon, Emu Software, Qlusters, Symbiot, Webmin, and Zenoss. The oldest of these is Jamie Cameron's Webmin, established in 1997, which provides an easy Web-based user interface for Unix system administration. The project is sponsored by OpenCountry, which joined the OMC in September 2006. The other founding members of the OMC also support free software projects, in a variety of ways. For example, Ayamon was founded by Ethan Galstad, who is the creator and lead developer of Nagios, an open source host and service monitor that uses a plug-in architecture to provide a rich range of options.

The case of Symbiot, which provides software for network security event and risk management, is more complex. The company was founded back in 2001, but initially sold only proprietary products. Then, as Symbiot's founder and CEO Mike Erwin explains: "We introduced an open toolkit and visualization platform called OpenSIMS in 2005, upon which a great degree of the Symbiot software is based. OpenSIMS is an independent package, maintained by Symbiot and programmed with hooks for other common open source packages." He says the benefits of this move flow both ways: "Open source code bases provide a method for end-users to do intelligent customization while providing the original code creators with [a] 'lighthouse' pointing them towards where the commercial space should go."

Emu Software took a similar path to openness. It started life back in 2003 selling NetDirector as a closed source Web-based system administration platform. "Although we always felt that we would contribute at least part of the product to the open source community," says co-founder Greg Wallace, "we concluded in late 2005 that systems management would be the next big computing market to see significant open source adoption, and we wanted to be out in front." He believes that certain sectors lend themselves to the open source approach: those where there are "lots of users; a horizontal nature - that is, cross-industry adoption; a high incidence of user desire to customize; an initial market dominated by large incumbent vendors with integrated, and some might say over-engineered, products."

Wallace explains how the OMC is trying to bring some order to the wealth of open source systems management solutions:

The collaboration efforts that I see as being most promising are those that will reduce the complexity for users of having multiple point management solutions in their compute environments. Having lots of point systems can be a huge headache, and it is one that some big vendors have addressed by building massive, integrated product suites. But these suites never do everything, and once users go down that road, they can become victim to lock-in. OMC promises a different solution: make our various systems talk to one another, and reuse as much of each other's architecture as possible. For example, one initiative that has been discussed is the concept of an open agent that would be shared by various systems. Were such an open agent to became ubiquitous, it would radically simplify systems management implementation, as well as make such systems far more flexible and adaptive, since users could leverage a common underlying agent architecture to turn on new management functionalities as needed.

And Erwin notes one practical benefit Symbiot has already derived since joining the consortium:

Our offerings sometimes rely on the collection or interpretation of data from other vendors. One such vendor is Nagios. Membership in the consortium has already given us great access to the key code committer (Mr. Galstad) which was invaluable in helping us set a developmental course.

Looking forward, Wallace hopes that the OMC will become "more structured, with some defined working groups and a more defined mission and by-laws. Eventually, I'd like it to function, and be organized, like Eclipse." Erwin believes its influence could be considerable: "In the long term, I see the OMC as being a central clearinghouse and repository for system management tools with not only the Big Four's participation, but likely guidance."

That may be some way off, but already the membership of OMC is swelling fast: just four months after its foundation, the original five members had grown to 29. Among them is Hyperic, another major player in this space, and with an interesting history. It was originally part of Covalent, which provides commercial support for Apache, before splitting off in March 2004. Like Symbiot and Emu Software, it too began selling closed source products before opening up its flagship software Hyperic HQ, a suite of inventory auto-discovery, monitoring, alerting and portal tools, in July 2006.

John Mark Walker, head of community development at Hyperic, explains the move: "From Hyperic's founding, it was always our intent to open source HQ - once we felt that it had reached a level of maturity to be useful for a number people, and once we had the in-house resources to properly support our community and foster its growth." And he points out: "The problem that existing management software strives to address - integrating with every existing and future technology in order to manage it - is only solvable through open source communities. It is impossible for a single company to keep up with all of the newly emerging software and other technologies in the data center. The problem requires the interactive, two-way communication inherent in the open source process.

Not everyone sees the OMC as the way to do this. For example, another leading company in this area, GroundWork, prefers to do its own integration of open source systems management tools to create its GroundWork Monitor product line, which includes both closed and proprietary elements. Although the company says it doesn't "see a particular need in being a part of the OMC at this time," it has created its own Open Source Council in August 2006, with the aim of ensuring that GroundWork "will always be comprised of the very best open source projects comprehensively integrated into a platform." Whether within or outside the context of the OMC, integration remains the key challenge for open source management tools.

Glyn Moody writes about open source at opendotdotdot.

(Log in to post comments)