We are in the middle of trying to determine how to do Availability and I was wondering how other people do theirs. The biggest discussion so far is around when to take an availability hit or how much of a hit to take when part of the service is not working. There is no list of critical business functions for each service and IT is going by what they think are critical business functions vs getting those from the business. My thoughts were determine critical business functions and as a whole they make up availability for the service. If one of those critical business functions is unavailable you would take an availability hit equal to the importance of the function. Any feedback on how you handle Availability would be great!
Joined: Oct 13, 2006 Posts: 116 Location: South Africa
Posted: Wed Sep 12, 2007 12:53 am Post subject:
Availability Management is one of the disciplines companies tend to deal with "last". And many do not progress beyond component availability (e.g. "SAP server 1 was 99.7% available last month").
When you deal with end-to-end service availability you have to consider how to average over multiple end-users. If one site out of 10 went down for 1% of the month, that's only a 0.1% hit on availability, ie 99.9% availability. (Assuming all the sites had the same number of users. The same number and profile of users, even.)
But you were asking about multiple features within the service. I think you should identify CBFs in partnership with the users - but when you look at availability, only worry about the ones that are likely to fail independently, and which you can actually measure.
Then you'll need to weight the features. If they don't rely on each other you could have CBF1 = 40%, CBF2 = 40% and remaining features = 20% - again, agree this with the users. The availability figures have to be meaningful to them.
On the other hand if they can't do anything without CBF1, then a failure of CBF1 is a 100% hit - 0% availability for the duration.
Also bear in mind time factors - if the service is only used during office hours, don't report availability for other hours. Consider critical business periods e.g. month end, and be prepared to report on bad cases - a day with ten 5-minute outages will definitely be remembered by the users even if your availability for the month is 99.5%.
Hi, I agree with JoePearson, and I would like to add another point.
Once you have identified your CBFs, you also need to determine all the components that make up the service and determine their importance. Some equipment could have failover or redundancy built in. You need to identify all of these components, and gather the total cost of maintaining that service at that level. This information is then used in your reports back to the business. It is these hard numbers of both the availability and cost that will give you support when the business want better availability. You can tell them it will cost an additional $xxxxx, for an increase of x.xx% in availability. Using the combination of these numbers is very convincing when presented to the business.
Availability is something that most people donít care about, until it is not there.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum