The IT Skeptic - Comments for "Somebody help me: what is the point of standalone CMDB tools?"

Asking for help?

cotswolddave — Tue, 13 Mar 2007 20:46:34 +0000

There is some truth to that, why is it that a buildings project manager does not ignore the advice of an architect, quantity surveyor, supplier of air con etc. I think there are many reason why implementations fail - customers often buy badly and suppliers won't highlight problem areas unless they have to. So the short term needs to manage incidents, changes etc. are focussed on while the tricky subject of config management is avoided.

While we use the term CMDB as part of the product name, in practice we hardly mention it when we are demoing or implementing. Its easier when talking to a customer team to focus on specifics such as a service map, or a list of servers that are critical to a process, or identifying where the test/dev/live version of an app is hosted. But with a marketing hat on - stick CMDB in the title of a seminar and you will get hundreds coming to see if the holy grail exists.

The CMDB concept is still worth striving for, becuase what else is there that attempts to understand how multiple technologies deliver services?

Have enjoyed the challenges and maybe evidence of one of our implementations will be leaked to the public.

reflects on vendors' services and support more than products

skeptic — Tue, 13 Mar 2007 19:55:45 +0000

Hi Dave

It sounds like what you are doing is useful stuff. Whether it is what ITIL defines as a CMDB is still to be argued, but it sounds closer than some.

I can't resist one last ex-vendor observation: the fact that clients of the Fat Four cannot get what they want often reflects on those vendors' services and support more than their products. Many of the product implementations I have seen exploited only part of the product's capability and then the implementation was often poor or just plain wrong bacause the client was left to sink or swim. There is a cynical merry-go-round where they buy one, then four years later decide it doesn't work and buy another. Either would have worked for them if (a) they invested and (b) the vendor cared.

Linking service maps to incidents/changes etc.

cotswolddave — Tue, 13 Mar 2007 13:23:07 +0000

Can't get one past the skeptic... He has too much experience and insight to believe my few words and is "skeptical"

1) In practice, if CIs are already linked to incidents and changes in the service desk then it is easy, using the CI ID as a reference. Our typical approach is to put changes, incidents, problems under the specific CI icon and have the item go red, flash etc. to reflect status, changes within last few days or planned etc. So that you can "see" where there are potential conflicts or lazy admin and things haven't been closed.

If you have don't have CIs and incidents/changes in the same system then you have a control problem and require manual checking. You have to use names as unique references for the linkages. For example an outsourcer and their customer can have two service desks, have different change processes but still understand and communicate on service issues.

2) Q. How do you reflect the current status when you are refreshing views? Well it depends. In a simple model you load up a service map and press refresh. If you have 100s or 1000s of service maps it is easier to do a timed automated refresh so that "hotspots" are noticed. As we may be generating all sorts of perspectives from a common source then you choose if there is value in having something updated within a day, week, quarter etc.

Many of customers see the production of a defined, owned, validated service map for every service as being the most valuable deliverable. Second to that is extras such as showing additional information such as changes, status etc. To put this in context, we are talking of projects lasting maybe a week or two for environments with 1000s of servers.

3)GUI Relationship Walkers
Not familiar with them although you could mean federated, dashboard or other terms where a view combines multiple data sources. It is normally their service desk systems that we are linking into - so there are customers who can't get what they want. If you or anyone else can give a lead on where to get info I'd appreciate it because there is too many concepts and little meat.

I suppose what we do is like the London A-Z. We try and link the index of streets at the back to high-level, low level and specific need maps (underground, congestion zone, hospitals etc) where a visual context may the best way to understand.

Still skeptical

dool — Tue, 13 Mar 2007 12:09:42 +0000

I still don't follow your line of reasoning.

Of all the big-name vendors, johnny-come-lately or not, only two won the rights to author a book (IBM is conspicuously absent). The remaining three books appear to be small European-based firms with authors involved since ITIL version 1. One must concede they "...have been behind ITIL for many years, genuinely are passionate about it."

So there we have it. A large pool of respondents, all passionate enough to compete openly and surrender IP, with subsequent winners representing old-hands and newcomers alike.

"I didn't compete because I didn't want to surrender IP," sounds like self-important posturing at best. A bit like saying, "I didn't compete for the olympics because I didn't want to reveal my running technique."

There you have it

JamesFinister — Tue, 13 Mar 2007 09:52:33 +0000

Ah, but there is a world of difference, perhaps, between those vendors who have been behind ITIL for many years, genuinely are passionate about it, and mature in their view of the market and those who are busy jumping on the bandwagon. Some of the latter, I suspect, are jealously guarding IP that we regarded as outdated ten years ago. I have to say that I haven't seen anything in ITIl 3 yet which appears to indicate there has been a massive injection of IP. The discusssions about CMDB and service catalogue going on in this blog indicate the books still have a way to go to catch up with the best real life thinking going on.

I don't buy the IP argument

skeptic — Mon, 12 Mar 2007 19:25:15 +0000

I'm with Dool. I don't buy the IP argument. Anyone who held back over IP is not thinking strategically and has a twenty year old view of how IP can be managed. They deserve to have missed out. As I said in an earlier comment they are getting paid while at the same time getting priceless kudos, exposure, knowledge, experience and contacts, and shaping ITIL in their own image. I'd crawl over broken glass for a gig like that.

Baffling

dool — Mon, 12 Mar 2007 13:26:38 +0000

I seem to recall the OGC website saying over 40+ vendors (many with multiple proposals) bidding for the opportunity to write a V3 book. I have to imagine all the usual big-name suspects were in attendance with their best thinking.

Why weren't they afraid of losing IP rights? They are typically the toughest defenders of proprietary IP. And what does that really mean? OGC will own their diagrams or descriptive terms? These things have very short shelf-lives to begin with.

If something is to be considered an industry practice, I imagine it has to be "battle-field tested" or at the very minimum peer-reviewed. Any good idea, particularly in a knowledge industry, is notoriously difficult keep from spreading. Sort of like sharing fire, everyone receives warmth but the new fires no longer belong to the original owner (forgive me for mangling an argument by Thomas Jefferson on IP rights).

Is it licensing? In many ways, people use the IP or take the training only because of the ITIL brand. Larger audiences mean that there would be increased opportunities and credentials for the author of the material - despite no longer owning the IP.

Or are we are discussing closely guarded closet systems with carefully constructed home-made process models? These models so good, they cannot be shared with the outside world? Well then, I think they made the right choice.

Sources in v3

JamesFinister — Mon, 12 Mar 2007 10:45:47 +0000

Hmm, well large parts of the volume I QA'd seemed to consist of summarised chunks of well known management texts. Though this was at least acknowledged I always think it is better to send people back to the original source. I know that several possible contributors to ITIL 3 held back because of the fear of IP rights being lost.

Overlap

JamesFinister — Mon, 12 Mar 2007 10:35:42 +0000

Funnily that scope list looks quite close to a the critieria I agreed a few years ago in an energy company for deciding the priority of incidents (though with the addition of damaging PR!)

Please remember that i was a software vendor for over a decade

skeptic — Sun, 11 Mar 2007 21:27:45 +0000

Please remember that I was a software vendor for over a decade. So when I hear "Its a fairly trivial exercise to then bring in incidents, changes etc" my radar goes off :-D

1) I'm assuming you mean that the CIs were already linked to the incidents and changes in the underlying service desk tool? Otherwise bringing them in would be highly non-trivial.

And 2) since incidents and changes are transactional, not static, how do you 'bring them in" frequently enough to be current?

Finally 3) I've seen GUI relationship-walkers in a few "cmdb" products. The Fat Four (IBM, HP, CA, BMC) all(?) claim to have import tools to 'bring in' data from other systems too. How does what you are describing differ from the facilities in the existing tools other than being cooler or better? i.e. what distinct business advanrtage does it give to justify the investment of the extra layer? Have i missed soemthing?

well it can de done!

Visitor — Sun, 11 Mar 2007 21:04:53 +0000

Hi Skeptic,

Maybe I should elaborate, you can draw service maps from a CMDB using the data that is in it. We do not attempt to "discover" relationships using any fancy technology because the world is too complex for such tools and the way services are categorised is manual anyway. There lots of discovery tools already out there but they are often technology centric (server,network, desktop etc) and they only discover technical data, none of the business or service stuff.

What do I mean by drawing service maps.

a. select a service, analyse all relationships with underlying components or CIs and then generate a picture which you can modify to make readable, embellish etc.
b. select a CI such as a server, analyse relationships above it so you can see software, systems, services and business processes impacted (harder to do than a.)
c. Select something in the middle like a piece of software and then show all supporting components and all services etc. impacted in a service map
d. Select a seed point such as a service and then show all relationships between shared components, services and processes in a service map

So you can rest easy in that there isn't a claim to have a magic bullet than creates a CMDB of CIs and relationships. But those who have expended the effort in using their service desk as a CMDB often find they can't use it effectively. Our CMDB analysis software just uses what they already have. The biggest benefit to be honest is to quickly validate how it is set up so at least it is a reasonable representation of the current state.

Its a fairly trivial exercise to then bring in incidents, changes etc. and show them on the same service maps so that you can see what or where CIs are affected. It works well in larger environments with 1000s of servers where you need maps to comprehend issues across the services. It does mean that capability to generate 1000s of service maps is necessary

But you do need the data to start with.... before you can analyse it.

discovering business process

ianclayton — Sun, 11 Mar 2007 08:11:46 +0000

we once had a situation where a business fax turned out to be a critical component of an order entry system, responsible for receiving orders from folks in the field. It broke. It was not regarded as an IT asset or configuration item. It was not discoverable. Until we walked the process with the customer and manually discovered it was truly a critical element. So when the fax ribbon broke and orders came out blank - money was lost. The true expanse of a service can be very, very suprising.

The etiquette of referencing sources

ianclayton — Sat, 10 Mar 2007 22:49:37 +0000

Well we shall never know if anything was gleamed from CMII - ITIL has a habit of NOT recognizing sources - I can offer CFIA (IBM), FTA (Bell Telephone), and Service Lifecycle, Service Provision Lifecycle, Incident Lifecycle and Holistic Service Management (ITSMI) as just the tip of the iceberg.... you can be sure we shall be looking for any such examples in V3...

Institute of Configuration Management

skeptic — Sat, 10 Mar 2007 22:42:25 +0000

Funny you should mention that. just last week I met someone who is CM-II certified - I'd not heard of it before.

I assume this is where some of that part of the ITIL IP came from??

There is a great overlap with ITIL. Seems to me the big difference is that CM-II is not service oriented. "The scope of CM, per CMII, includes any information that could impact safety, quality, schedule, cost, profit or the environment."

Who defined a CMDB first?

ianclayton — Sat, 10 Mar 2007 22:03:10 +0000

You might find that ITIL was not the first to define the concept CMDB - check out the Institute of Configuration Management - also check out the term/concepts related to Metadata... we need to remind ourselves of the objectives/needs that result in the call for a CMDB type solution. Relate all the elements needed to provide a specific service and the service level objectives. Thsi should take you directly to teh activities performed by the users - and then use those to map relevant infrastructure - ANY item - whether owned by IT or not - and you thought just the IT components were difficult to find and relate!

V3 catching up...?

ianclayton — Sat, 10 Mar 2007 21:59:16 +0000

We have distinguished service assets from plan infrastructure and customer assets for years and it is reflected in our ITSM Body of Knowledge (ITSMBOK) in the form of service asset management and service infrastructure management practices. We take a truly holistic view to try and avoid these issues that are largely caused by a perpetual IT view of service. Its encouraging to hear that ITIL V3 respects this and is effectively catching up to the real word experiences. I have not been able to review any V3 materials yet - any more insights as to how they define a service.... did they go as far as recognizing that 'vital business activities' - or those that drive results - are teh main service elements - you must check out our ITSMBOK!

Nice points

dool — Sat, 10 Mar 2007 21:45:18 +0000

Your point about the fax reminded me of a distinguishing term I encountered during the ITIL Service Strategy Public QA. The draft made a point to speak to "Customer Assets" as well as "Service Assets." Customer Assets are tightly coupled to Service Assets but frequently overlooked. A customer asset can be a fax, business process or person.

Without Customer Assets, services have little point. A simple but, as you point out, essential observation. On second thought, it is a brilliant observation precisely because it is so easy to overlook or consider irrelevant.

Service Assets affect the performance potential of services which in turn affect the performance potential of Customer Assets which in turn affect the potential of business outcomes.

V3 grows more and more interesting every day.

Love this point. A

skeptic — Sat, 10 Mar 2007 20:50:34 +0000

Love this point. A fax!

CMDBs cannot discover everything, and tools that discover are not CMDBs

measuring a service as a black box

skeptic — Sat, 10 Mar 2007 15:07:54 +0000

Maverick, as you say measuring a service as a black box is possible.

Not only is it possible but it is essential. See my latest blog entry (thanks for triggering it)

Dave, I’m afraid I too

BrendanM — Sat, 10 Mar 2007 12:11:17 +0000

Dave, I’m afraid I too remain unconvinced on this one, but I am strongly hoping I can be presented with scientific fact to back it up. I had been looking at a few tools which touted the capability. One is an upcoming product from Citrix, which provides for alerting and measurement of the end user experience within the MetaFrame farm, and the other was a Blue coloured Performance Test manager. Supposedly able to measure and track the responsiveness from and an end user perspective, a Java Web based application server (Virtualised) using a range of external middleware services connecting through to an enterprise database. (Mainframe)

Lets look at the equation, end user to MetaFrame, MF server operability (Sessions hosted, concurrent processes, page file swapping r/w, Concurrent library of applications with multiple backend dependencies causing undue processing (incl Domain chatter, print network, ACL’s), Network Latency, WAN Traffic Prioritisation, etc. Java Web Server fronted with Proprietary Web Platform Load Balancer and clustered set (Host in VM Farm), with service bus, middleware layers exporting and loading into several enterprise DB’s in a mainframe environment and the vendor say “Hey No Problem”.

I do know that it is feasible to monitor these components and draw analysis but to auto discover and construct the service map may be somewhat difficult. I personally recommend a manual approach to mapping these services with a little technical tool assistance. Once defined they could be loaded into the Open Source Monitoring framework Nagios under Orion config, and detail all services and hosts under a child / parent / grandparent relationship to enable ongoing monitoring. There is also a Nagios plugin called Webinject that is able to perform the initial call to the web server, initiate DB access, DB Commit, Transaction completion, and report the round trip at whatever preset service timing check you specify.ie 5 mins.

I look forward to being convinced it is mature and accurate enough to be presented to enterprise as a qualified service worthy of the funds expended for the engagement.

Please drop me a note at one of the cmdb.info forums if you want to discuss the technical attributes, Thanks and look forward to hearing from you.