Back in early 2004, Data Center Markup Language (DCML) was all the rage. Industry press predicted DCML would revolutionize...
data center management by providing a tool to manage data center chaos. Since then, there has been little news about DCML. Is there really something to it, or is DCML just another overhyped industry standard ultimately doomed to obscurity?
Origins of DCML
DCML got its start when a group of vendors, including Plano, Texas-based Electronic Data Systems Corp. (EDS) and Opsware Inc., of Sunnyvale, Calif., tackled the longstanding problem of managing data center environments along with automatic server provisioning.
"We came up with the assumption that it probably made sense to have a standard way in which elements are managed in the environment," said Darrel Thomas, chief technologist of hosting services at EDS.
The group put together a set of papers and strategy documents; then, it began talking to various industry leaders, ultimately deciding to create a 501C nonprofit organization to foster the standard's development. The resulting DCML Organization released draft 1.0 of the framework specification May 24, 2004, before moving the standard to OASIS last summer. The group's sponsoring members currently include EDS, Opsware, Tibco Software Inc., Computer Associates International Inc. and BEA Systems Inc., among others.
"The concept was to take knowledge developed at the element level, add a thin layer or import a layer of intelligence between elements, then make that kind of the grammar for a standardized way to look at configurations in IT environments," Thomas said.
DCML's strengths and weaknesses
Nearly everyone agrees that DCML fulfills, or at least attempts to fulfill, a critical market need.
"With growing interest in utility computing and varying vendor applications of that concept, there was a need to create some operating standards for how various technologies could fit together," said Jeff Kaplan, a former industry analyst and founder of IT consulting firm THINKstrategies.
Thomas hopes DCML will ultimately bring a number of different benefits, including enabling automation and providing a level of inherent business continuity. Being able to understand blueprints in an environment would allow companies moving to another facility or carrying data across a data center to rapidly reconfigure a catastrophic environment without mass upheaval, he explained. Previously, most means of accomplishing this involved falling back on proprietary software.
"Establishing certain industry-accepted standards closes the gaps between various proprietary implementations," Kaplan said.
However, not everyone believes DCML is up to the task it sets out to tackle. Charles Betz, a prominent technical consultant in Minneapolis, and publisher of the erp4it weblog (www.erp4it.com), thinks DCML means well but may be fatally flawed in its structure.
"They are attempting a metamodel using Semantic Web standards that are not baked yet, are impenetrable by mere mortals and are not supported by any talent pool or serious production technology," he said. "For a basic information model challenge like DCML bounded by the walls of an enterprise's data center, it is complete overkill."
Where DCML is headed
For believers, DCML's biggest challenge is clear. While the standard has several prominent vendors behind it, DCML has yet to garner support from the largest names in the data center space. For example, companies like IBM, Hewlett-Packard Co., Microsoft and Cisco Systems Inc., among others, are noticeably absent.
Betz said he believes larger companies may be biding their time while the market is still so immature, and they may not see DCML and related efforts as a revenue builder. Additionally, he forecasts DCML will eventually face competition from another source.
"I think it's much more likely that the OMG/DMTF [Object Management Group/Distributed Management Task Force] alliance will succeed, or perhaps some purely proprietary play by IBM or HP," Betz said. "Another possibility is the proliferation of built-for-purpose XML schemas requiring point-to-point mapping."
Kaplan chalks up the absence of large vendors to politics. "It's obvious this effort was being led by EDS, which was a competitive disadvantage to IBM and HP," he said. Kaplan predicts the standard will remain stagnant until more companies come on board and until the DCML work group begins to pump out frameworks that will demonstrate changeable value.
According to Thomas, EDS and DCML's other backers are currently focused on recruiting additional member companies and getting the word out about DCML, as well as creating and remapping collateral and presentation material. Thomas said he believes these efforts will soon start to resolve the problem of large vendor participation. He predicts consumers will compel companies to support the specification, and he expects to see DCML introduced to product sets by early next year or mid-quarter next year.
"The process framework strategy is a good one. It relates DCML usage to line of sight and to the holistic level of customers' existing environments," he said. "Any big company trying to standardize its environment is going to be innately interested."