I am having an internal discussion in my head I could use some input on.
It is standard in DCRUM 12.1 that TCP connectivity errors are what's included in the calculation for the metric "Availability".
I want to change this to "All TCP errors", but how will this affect the Availability metric?
I am not sure if it will use the metric TCP errors to then calculate Availability, or if it will include ALL TCP errors, which would include errors of the type "server session termination errors". In this particular case there is applications using this kind of RESET packets to close TCP sessions (unhealthy and old fashined, but true), so it would completely mess up the statistics if it is included in the Availability metric. Does anyone have any input on this?
I think that depends on what is included in "All TCP errors" and I can't really tell as I can't find any explicit reference about that.
If for example checksums or retransimissions are considered an error, you might end up with every packet being classified as an error depending on the source of the traffic. I would recommend opening a ticket to get the labs view on what is classified as an error. Please share you findings here
I know, there is no reference to this challenge, and thus it would be nice if someone had some experience with it.
I did open a ticket on it prior to posting here actually. I can share what ever they come up with of an answer. Note that this is only a problem in 12.1, because the Availability metrics are changed in later versions.