A cool look at ITIL

Blog entry submitted by skeptic on Mon, 2006-05-15 22:01. [nid:1]
in

Now that ITIL is the de facto standard for IT operations, the time is ripe for a more objective evaluation of ITIL’s merits and caveats. Let's do that on this website. In the ITIL world it is still spring or summer.

This blog seeks to balance that with an icy blast of winter through the techniques of the skeptic – consider the observable facts and question the underlying assumptions – as well as applying that other great Litmus test: common sense [Common sense is something that used to be common, hence the name. You youngsters look it up on Wikipedia].

The IT world is traditionally split into development and operations halves. In the operations hemisphere, ITIL has been the centre of attention for most of this century. If you have anything to do with running computers and you haven’t heard of ITIL, you should read up on it, if only to be in the know at those awful parties where IT people talk their own language like some secret society.

Gartner have a most useful model for considering the waves of irrational exuberance that regularly sweep across the IT industry: the hype cycle (see a picture of it and more info here). ITIL is somewhere around the peak, though it varies around the world. I think it is not in the trough yet anywhere: it is still greeted with acclaim and enthusiasm and often inflated expectations. But progress down the slippery slope is beginning. Hopefully a little objectivity now can reduce the height of the peak and the depth of the trough, and ease the transition into a more stable maturity.

Please come along with me on this journey as we explore the winter side of ITIL. I will post more thoughts approximately every when I get around to it. Your comments are not just invited, they are expected. If we get enough interest, I will open up a forum here too. If anyone feels they have enough material on this topic to run their own blog, contact me.

I urge you to please consider registering as (a) you can then get notifications when I add something (b) you can take part in the forum if we get one later and (c) it's only polite to identify yourself. You can still add comments even if you remain anonymous: I'm more interested in your opinion than your identity.

[There are many more reads for this page than the others in the blog. You guys coming direct to this URL know there is more to the blog eh? See the blog link below or in the menu at top right]

OK look away now, go read the rest of the blog. This next bit is for the search engines:
Still reading? You would be amazed how many people search on "[something] sucks" to find or measure negative sentiment, so let's say ITIL sucks even though I don't think ITIL sucks and i wouldn't say ITIL sucks but just for those who might have used ITIL sucks as keywords. Likewise people may look for an IT critic. I don't mind being called the IT Critic, though IT sceptic is more accurate I think. Though it is fair to say I am being critical of ITIL so IT critic fits. And we do cover ITIL problems here because there are problems with ITIL and a sceptic would certainly examine ITIL problems. Note I spelt it IT sceptic for all those good folk who use the dictionary spelling of sceptic not the skeptic variant we skeptics use.

Bookmark this post with:
  • Delicious
  • Digg
  • Google
  • StumbleUpon
  • Yahoo
  • Reddit
  • folkd
  • Facebook
  • Propeller
  • Magnoliacom
  • Newsvine
  • Furl
  • Rojo
  • Technorati
  • Icerocket

Comments

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Configuration Item

Submitted by stapelt on Mon, 2006-10-30 03:13.

What I like to know is exactly what is the definition of a CI? How does one define the CI for a physical server for instance. ? Is the DNS name of the server an attribute of the server or the primary object name? when the servers memory is upgraded or the operating system is upgraded is a new CI created? Can anyone tell me how we manage the CIs when a server is replaced with a new box completely -- is this a new CI and does it have a historical tracking to the older technology box?

what is the definition of a CI?

Submitted by skeptic on Mon, 2006-10-30 05:31.

The ITIL meta-data sucks. There is no good definition within the ITIl materials of what the data should look like (the best you will find is on page 164 of the blue book). So "officially" there is no answer to your question. As a result there is much debate.

Is the DNS name of the server an attribute of the server or the primary object name? Yes. Whatever. Your call. I would say that service management is about user-oriented computing, so using the DNS as the primary object name is unlikely to be meaningful to many users. others will argue that doesn't matter as the user ses the service not the underlying CIs. More debate.

when the servers memory is upgraded or the operating system is upgraded is a new CI created? No, the properties of the server have changed. Unless of course the memory in the server is itself a CI, which is entirely permissible - but daft.

when a server is replaced with a new box completely -- is this a new CI? Probably, because among other things the CMDB should support Financial Management

does it have a historical tracking to the older technology box? Yes, because a CMDB maintains checkpoints over time. There is a physical CI which is the server ("HP Unix box A12745") and there is a logical CI which is the function the server performs ("webserver #1"). The current version of the CMDB relates the new box to the function. A historical checkpoint of the CMDB relates the old box to the same function.

I'm amused you ask the question here, as this site argues that the whole CMDB concept is a crock of the proverbial. You will get a wider range of opinions (which is all there is on CMDB design) here or here

CMDB and CIs gone amuck

Submitted by stapelt on Mon, 2006-10-30 14:39.

My questions were meant to be somewhat cynical from the standpoint that they are not answered and I know that. I have worked with a major vendors product for 3 years now. I some deep knowledge of the datamodel that is used. I can tell you for sure that when it comes to best practices ( what ever tj\hat may mean ) in this case decades of modeling practices have been thrown ignored. But this leads back to what is the CI name, because in this models case even the main name of the CI does not have a unique key constraint. In fact the entire model is held together with surrogate keys ( Object Identifiers they are call ). This design 'feature' has been carried to the extreme where almost 50% of the fields in the system are object identifiers.
The result:
1: poor performance that will be difficult or impossible to fix with hardware. For DBs are less than a million records or even in the hundred thousands we should not have performance problems.
2: not scalable
3: the user interface is horribly difficult to use -- except for the black belts
4: The reporting capabilities of the system is terrible complex and only highly trained employees can extract the information. Also because the system uses weakly typed data the usability of reports is undermined severely.

ITIL metadata is a mess.

Submitted by skeptic on Mon, 2006-10-30 21:24.

Duh. Sorry, I missed the tone. That's the trouble with email and forums: without voice and body language I for one get the wrong end of the stick.

I suspect you work with the same product I did, or two vendors are making the same mistakes which is entirely possible.

ITIL metadata is a mess.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.