Does the Server Availability Index (Show ai) and the resulting server_transinfo_range matter on a non clustered server? I have a server that goes to availability of zero all day long but it is not clustered. Would it make any difference to set server_transinfo_range on a non clustered server, or for that matter does server availability even matter on a non clustered server?
Subject: clustering aside, it is a load indicator
The fact that the help document begins by talking about clustered servers is misleading. Server Availability is a measurement of that server’s response times as measured against ITSELF. (That’s why when you have a brand new server, you can’t rely on the server availability index to give you anything useful.)
When it has a lower availability number, that means it is slowing down - usually from being loaded down with users, tasks, addons, etc.
While this figure can be used to balance workload in a cluster, it can ALSO be used by an administrator to keep tabs on NON-clustered servers.
It doesn’t make much sense to manually change the server_transinfo_range in a non-clustered environment, as the primary use of having this indicate a fully loaded server at a certain value would be for failover to another cluster member. In a cluster, you might change it because different servers are more/less powerful…30% of a very powerful server is faster than 30% of a slow server, so as the Help says: “If your server is particularly powerful and fast, you might want to increase the value of the expansion factor that is considered fully loaded.”
Subject: a relative performance metric
I think it’s still meant to be a handy reference for how heavily loaded your server is over time. If you set the SERVER_TRANSINFO_RANGE parameter according to their recommended level from telling it “show ai” after a peak load and it’s going down to 0, that might mean you have an issue.
maybe this tech article applies?
http://www-01.ibm.com/support/docview.wss?rs=0&uid=swg21142084
Subject: It does matter
Hi Sean,
Clustered or not the AI matters. A poorly performing server will have a lower AI. Quiet often if this is a Wintel box it is down to the lack of recent reboot. The memory leak issue on Wintel boxes means that less and less resources are available or the server to perform at its optimal. It is a measure of itself against itself over a period of time.
It could be that the server is being overloaded with more and more tasks or that some tasks are running and not completing in time and holding up rescourses also.
It is just an indication of the current health of the server.
Hope this helps
Paul