Discussion created by Chris548 on Feb 27, 2013
Latest reply on Feb 27, 2013 by 1_keithk

Does anyone have a lot of experience with SLA's?  If so, I have a question around SLO calculations..  I have an SLA set up with one SLO.  That SLO has only one QoS object and it's informix availability using the check_db_alive checkpoint.  The SLO has different calculation methods and is set to average by defualt.  The other night we had an issue with a database that prevented the probe from running correctly and running its check_db_alive query and the probe just timed out each time.  So there isn't any null data in the database, just missed times I guess.  If I change the calculation method from average (which had been giving me 100% availability which isn't true at all due to the outage) to sequential, I notice that the SLO is now achieving 98.31%.  Is that because it can tell when the QoS intervals are not in sequence and it can see that there are missing times?  The documentation doesn't really indicate that's what sequential is at all.