SLAs during contingencies

If we combine the SLA concept with the contingency plan concept, the first thing that will occur to us is the service level agreement that we have concluded for implementation of the plan, or, in other words, to restore functioning of the services affected by the contingency. But there is actually more to this: What happens in the case of services that continue to operate normally? Should they continue with the same level of service? Well, this depends...

Imagine that we have a service level agreement with our users stipulating that a request for service to substitute a mobile terminal must be resolved in a maximum period of 16 hours (a couple working days). But now suppose that an emergency has arisen at our main DPC and that the Active Directory is out of service, along with the email, ERP, and CRM... The situation is not very promising, and it is clear that we are faced with a serious contingency, and that now we will have to pull out all the stops to restore the services affected ASAP (or at least within the agreed period). And during this kind of emergency, does it make sense to maintain the delivery period for terminals in 16 hours? We might think that if this has been agreed, then we must abide by this time frame, but...

• It will be of little use to have a terminal if we cannot provide the rest of the services necessary for normal use of this terminal.
• Perhaps we are taking up resources that could successfully reestablish the unavailable services

Wouldn’t it make sense if, during a contingency, certain service level agreements could be loosened somewhat? (at least those regarding non-critical service requests).

Defining the SLAs

Obviously, we can’t do this unilaterally: the relaxation of the agreements must be arranged with all the relevant players, but it is one more factor to take into account when defining the SLAs:

• What are the SLAs during “regular operation”?
• What are the alternative SLAs during a contingency?
• Which SLAs must remain unchanged regardless of what happens?
• And of course, what is the definition of a contingency? And we must take into account that certain circumstances could be regarding a disaster for IT, but be completely irrelevant (or even imperceptible) to the end user.

I hope that you enjoy it

Jandro Castro