IT Management and Cloud Blog

« | Home | »

5 Questions for the ITSkeptic

By John | October 29, 2008

The StackSafe guys starting asking certain bloggers questions and then posting them on their site. I thought that would be a great idea to ask some of my favorite bloggers five questions and if they were interested they could ask me five questions. This is my first crack at it. However, I may have shot myself in the foot by exposing how much better my favorite bloggers are at answering questions then I am. Here is a link to the five question I have answered on the Skeptics blog.  Rob England (The IT Skeptic) brilliant as always…

1) Last year I posted a question on the Tivoli forum asking “Does ITIL really matter?”. I asked this question as it relates to the new kind of “Google” type data center. Some of these new enterprises design their infrastructure on cheap commodity hardware and expect failures as way of normal operation. So I ask you “Does ITIL really matter ” for the new commodity type data centers?

Frankly I don’t understand the question. In what way do the new data centres differ from the old ones that would make a difference to ITIL?

Most of the objection to “ITIL” from operational people seems to be over Change Management.

First, IT Service Management is about making the service delivered to the end user the first priority instead of technology for technology’s sake. I have to say I don’t have much time for all the disgruntled geeks on that thread howling about how bureacracy gets in the way of their genius for operating machines. They insist on operating in an uncontrolled environment and we are supposed to accept the one day they cock up and take a few hundred or thousand users out as the price we pay for their unfettered brilliance on the other thousand days. I don’t buy that. People make mistakes. IT is becoming a profession, which means behaving professionally, which means operating within whatever management constraints ensure maximum benefit to the people paying for us. Get over it. It also means treating our knowledge of the environment we manage as the employer’s property not our personal feifdom.

Second, if these newfangled environments are so robust and self-healing that cowboy cockups don’t take out a service, then fine, those particular types of changes can be uncontrolled. Nothing in ITIL to say they can’t be. My policy with change management is always that IF the geeks can show that such and such a class of change has happened so many times before without failure or with seamless recovery, and if that class of change can NEVER trip up some more complex change inadvertently going on at the same time, and if it is documented so a new boy can do the same and can back it out, and if the risk of the inevitable balls-up anyway is low enough, then by all means make it a standard (read: pre-approved) change. But nothing exempts changes from being recorded. If it all ends in tears, other people need to be able to find out what changed and they need to do so quickly. Even if the changes are being made by the system not by a person, I think they should be recorded somewhere as a change. Otherwise how can we possibly know what the current system configuration is?

All of my ugly experiences with service interruptions that I can think of have been either operations or development making a supposedly simple change.

Ultimately though, ITIL doesn’t give a flying fox how the data centre works. Very little that ITIL cares about is influenced by that, other than some aspects of changes, availability and capacity. The cloud is just one layer of implementation of a service, and no matter how wonderfully infallible it is there are many more links in the chain before a service arrives at a user. That service still needs to be presented to its consumers at an agreed level, and we need to be accountable, and to monitor and report on the levels of delivery. Requests, interruptions and changes still need to be known and dealt with. Problems still need to be hunted down. The costs need to be managed and allocated. Services need to be invented, planned, tested, implemented, evaluated and retired. What’s any of that got to do with how the datacentre works, or did I miss something?

2) Where’s the beef with open source and ITSMF?

ITSM is not an exciting topic to most folk. Those of us attracted to it tend to be conservative by nature. To be blunt, there is a high proportion of old codgers in itSMF. Add to that the fact that ITIL is owned by the British Government and you have a formula for some serious conservatism. It shows in things like open source, or more precisely open content, as well as in collaboration, community and all that left-wing hairy-armpit Web 2.0 stuff.

Microsoft just put MOF out under the Creative Commons licence. It’ll be a cold day in Hell when OGC do that with ITIL.

Likewise, itSMF finally has an online forum. But there are plenty of established forums taking the traffic, and they do little or nothing to promote their forum. itSMF probably think this is kinda funky, but there is no collaboration mechanism for users to propose or contribute ITIL content or to rank or vote on the existing content. Online communities like Facebook and LinkedIn have thriving unofficial ITIL and itSMF groups. When itSMF realised, their response was to get heavy about use of the brand.

The long-awaited ITIL website with additional information about processes etc has finally arrived and guess waht? It is copyright, owned by the ITIL publisher, and costs thousands of dollars a year to subscribe to.

They just don’t get it.

There are people trying to drag ITIL and itSMF kicking and screaming into the 21st Century but they have a long road ahead.
<!–[if !supportLineBreakNewLine]–>

3) I’ll throw Drupal right back at you. What can Drupal do for ITSMF?

They already use it :-D

They could do more. itSMF International provides sod-all centralised services. Websites, membership, billing, booksales, emailing, elections, surveys – everything is done redundantly by fifty-something chapters around the world. Drupal Groups is perfect for allowing itSMF to host all the itSMF chapters centrally and provide all these services for them.
<!–[if !supportLineBreakNewLine]–>

4) In my experience auto discovery of a CMDB seems to be the last mile. What has been your experience of best practices as it relates to population and maintaining consistency of an enterprise CMDB.

I think autodiscovery is the easy bit. See, autodiscovery only finds the boring stuff, that which exists. All of the intangibles such as logical groupings into business processes or services can’t be autodiscovered. Nor can priority, agreed service levels, warranties, costs, or business owners. So all of that needs to be manually discovered, maintained and linked to all the physical stuff. And it doesn’t always get all the physical stuff, though fortunately most of it is IP connected these days. But UPS, aircon, physical security systems, PABX, embedded computing in machinery…. can all be a challenge to autodiscovery. Even mobile devices present their own challenges.

Autodiscovery helps with about half the initial population of the data, but it still needs to be checked.

A popular misconception is that auto-discovery can automatically keep the configuration data current after the initial population. it can’t. All changes should be happening via change control (see above) and part of that process should be to update the configuration, including working out and recording/updating all the conceptual data and relationships. We don’t update it by dumping uncontrolled autodiscovered data. The autodiscovered data is only useful to audit what is in the configuration to identify unauthorised changes.

For these reasons and others I think populating and maintaining an enterprise CMDB is only something that should even be attempted by a small proportion of the most complex and mission-critical IT installations. For most sites, the cost simply will not return the best bang for buck. Sites need CMDB like a car needs a GPS navigator. For a few it is justified (couriers, emergency services). For most it is geeks spending money for their own tech satisfaction.
<!–[if !supportLineBreakNewLine]–>

5) You had to know this one was coming. What are your thoughts about IBM Tivoli’s service management offerings as compared to other enterprise solutions?

The IT Skeptic tries to stay non-partisan, except when vendors do dumb stuff. So the Big Blue dropped a long way in my estimation when they sold TSD, and I gave them stick for that. I also think their recent return to the ITIL religion and attempt to recapture the high-ground lacks credibility and commitment.

Also technology bores we witless these days, so I tend not to know much about it. As you point out, Operations is becoming commoditised, and this includes enterprise management technologies. Far as I can see, they all pretty much work, near enough. A bell here, a whistle there. Ultimately all the commercial vendors will have to deal with the rising open source community. They’ll end up giving these tools away – what else can they do in the face of Zenoss and Nagios and Puppet etc? The new frontiers are the Cloud, which you track, and Service Management, which I do. The vendors are rushing there to find leading edge stuff they can sell at a premium. It will be fascinating to see if the vendors can stay ahead of the open source community.

Topics: itil | 2 Comments »

2 Responses to “5 Questions for the ITSkeptic”

  1. IT’s About Uptime - The StackSafe Blog » Blog Archive » Links List 10.31.08 Says:
    October 31st, 2008 at 1:12 pm

    [...] Willis in the past. It seems we also helped to set up a fine connection between the two as both posted brief Q&A’s with one another. Check it out for more information on ITIL, ITSMF, and what makes [...]

  2. Links List 10.31.08 | ScienceLogic Says:
    October 31st, 2008 at 4:11 pm

    [...] of our favorites, the IT Skeptic was featured on John Willis’ blog this week, answering some questions about CMDB, ITSMF and more. He also provided his insight into [...]