I attended the IT Service Management Forum USA Conference & Expo 2004 in Long Beach, California last week, taking in three full days of conference sessions, vendor presentations, and keynotes. It was a good conference; interesting presentations and lots of energy. IT Service Management, and the ITSMF as an organization, are evolving quickly and the sense of momentum was palpable.
I presented on “Metadata and IT Service Management,” and attracted probably 100 folks to hear what I had to say. While I don’t know what the majority thought, several people approached me after the talk to indicate they had found it relevant to problems they were struggling with in their practices. Advice to session attendees cautioning them against ill-considered data designs was received well, as was the recommendation to involve enterprise data architects as soon as possible in any CMDB effort.
Configuration management continues to be one of the hardest things to concretely execute in the ITSM/ITIL vision. I did a quick survey of the audience and found only one person out of 100 (a fellow from a major telecomm firm) who claimed to be effectively handling configuration management for all the below:
-Services
-Applications
-Servers
-Databases (including tables and columns)
-Queues
-ETL data feeds
A continuing theme discussed in several forums was the need to bring enterprise architects into the process whenever logical CIs are concerned. Logical concepts such as application, system, service, and process are the core concerns of enterprise architects, who are the best qualified staff in the enterprise for logical CI identification.
This is a disconnect in the ITSM world. For better or worse, ITSM practices tend to be driven by operations folks, far removed from enterprise architects. The risk is that those who claim ownership of ITSM processes will bypass their enterprise architecture colleagues and get bogged down in the complex conceptual issues that emerge in attempting logical analysis of IT services.
One ITSMF presenter suggested that the Federal Enterprise Architecture classification taxonomies be brought into Service Management tools, which would be a radical development; currently such taxonomies are only available in tools such as Popkin or Adaptive.
I'm hoping that the ITSMF starts to book folks like John Zachman for future conferences.
Vendor expo
The vendor expo provided further evidence that configuration management, and the concept of CMDB, continues to present challenges. There is no consensus on the scope or meaning of the term CMDB, for starters. Axios primarily manages desktops, providing no coverage of servers or even enterprise services, yet calls its database a CMDB; Troux on the other hand is more of an architecture tool and IT data mart, also calling its repository a CMDB. CA Unicenter looks more like Axios, while Remedy is readying what looks like a very strong tool. There were several vendors (such as Relicore and Cendura) providing standalone CMDBs; while there is a strong momentum for ITIL suites that incorporate multiple process areas, a case can be made that best-of-breed is a viable strategy for solving the configuration management challenge.
I won't embarrass the vendors by naming them, but the naive "any to any" CMDB is still alive and well; I saw at least one tool that would allow arbitrary linkage between any CI class -- a horrible idea. Other tools are much more mature, however; some clearly have been designed by real data architects.
Exhibitors list is here. Those who missed the expo were as interesting as those who made it. Notable by their absence:
-Opsware
-Peregrine
-Viadyne (who apparently have been purchased by Remedy, who are in turned owned by BMC).
-The DCML and DMTF standards bodies (very disappointing)
-Troux (had a rep there, but no booth)
-Mercury Interactive (with their recent purchase of Appilog, they should have been there)
-Altiris
Finally, one of the most interesting discussions I had was with a professor from Carnegie Mellon (home of the Software Engineering Institute). I said some rather provocative things to him re: the academic community's general lack of relevance to IT Service Management and the problems of running large scale enterprise IT, and he agreed wholeheartedly. I'll probably cover this more in a future article.
All for now,
Charlie
