Cash and Burn: Of Standards and 'Marketecture'

Virtualization is hot. That's no secret, but along with its ability to simplify and consolidate server operations, there lies an unexplored minefield that comes from adding any new layer of complexity to an existing system.

Today, most data centers contain a hodgepodge of equipment from different vendors, all of which come with proprietary management systems. Virtualization adds another challenge--namely, correlating what's happening on the virtual machines with what's occurring on the physical servers. Throw in a cool tool like VMware's VMotion, with its ability to move VMs at will, and you've got a management challenge that can't currently be met.

UXcomm hopes its SOMA (Service Oriented Management Architecture) will be the solution to managing the virtualized data center. SOMA creates a Web services interface for network and systems management. Instead of writing management agents to a weak interface--like SNMP, or worse, a proprietary one--third-party device vendors write to an open (in theory) Web services specification. Equipment manufacturers could then publish these SOMA-compliant WSDLs (Web Services Description Languages), enabling any SOMA-compliant console, such as UXcomm's XManager, to manage those devices. Such an architecture provides for heterogeneous systems management without the overhead, lock-in or a high-cost, high-end management platform.

UXcomm also hopes to be one of the first vendors to bridge the gap between physical server and VM management. With its recent acquisition of Virtugo, UXcomm has the pieces for managing VMware with Virtugo's VirtualSuite, and for managing physical hardware with its own XManager. A successful integration of the two products should provide capabilities not found in systems-management products.All this would seem to make UXcomm a vendor to watch. "It's a simple argument to make to CIOs," says Earl Hines, director of product marketing at UXcomm. "For every dollar you spend on virtualization, for 15 or 20 cents you can increase throughput by 50 to 60 percent." Sounds good, right? Simple to explain, cheap and great performance improvements; you can't ask for more than that.

On the face of it, I agree; it does sound great. An open, Web services-based architecture is exactly what this industry needs. Blending a WSDL interface with an understanding of virtual and physical infrastructure will undoubtedly simplify the management of tomorrow's virtual data center. Unfortunately, that's not what we're getting.

UXcomm's efforts would be far more credible if the company was working with competing vendors to develop an industrywide SOA definition for system and network management--but they aren't. SOMA, as it stands today, appears to be simply another marketing move to gain short-lived media attention. The interface carries no major industry backing, no standards committees, and misses on a number of key server-based interfaces, such as SMASH (Systems Management Architecture for Server Hardware).

SOMA also lacks the third-party developer support needed to form a coherent view of the entire virtualized environment. For example, UXcomm competitor Cassatt provides support for VLANs and enables the movement of server images between virtual LANs. To be effective, SOMA must provide a holistic view of the environment--and that requires VLAN support as well as awareness of the entire virtualized infrastructure, from storage systems to server farms, routers and switches. UXcomm can't go it alone; such a universal view requires the active participation of a broad vendor consortium or standards body.

There's no doubt that UXcomm is on to something. The idea is a good one, and using XML fundamentals to implement the exchange of systems management data is also the right way to go. However, it's equally certain that UXcomm doesn't have the influence or capacity to deliver all the pieces of a holistic management system. If UXcomm really wants to make an impact on systems management, it must turn over SOMA to a standards body and hope that enough major vendors become interested to achieve critical mass--otherwise it's just one more good idea that never made it much beyond the idea stage.Executive Editor David Greenfield can be reached at NetMagDave across every major IM system, or, if you must, via e-mail at [email protected].