In relation to RAS servers I am trying to understand on how does all this work ?
Currently we have Open connections set to 75 (0-180 is the min and max values).
So what are the implications if we increase the no. of connections above 75 for each RAS server say 90 or even 100 ? What impact does that have on the server side(memory wise or cpu core usage) ?
If we do not increase the no of connections and ADD another RAS server to the existing infrastructure then what are the implications on performance wise and which option is recommended and why ?
The current setup we have is
3 RAS servers (with 75 connections) per Unix box in a clustered environment so in total we have 6 RAS servers combined making to 450 connections.
We are getting 500 connection request so should we add another RAS server or should we increase no. of connections under each server to 90 or 100.
Please advise asap..
This will be greatly appreciated.
Can you also point out
For RAS, is the connection from CMS kept open until 1) Until the report is generated? 2) Until the report is read and the WebSphere/CMS connection shuts down
For Oracle, is the DB Connection kept open by the RAS 1) Until the queries are done? 2) Until the report is generated? 3) Until the CMS connection to the RAS shuts down?
In some scenario’s, if RAS connections are > 100, there are more RAS connection capability than CMS allows, thus we need to know if we can have multiple CMS on the same systemor is does a CMS handles only upto 500-600 concurrent connections ?
Our current configuration of servers is :-
6 x SUN UltraSPARC VII+ CPU cores
72 GB local disk space
NAS DISK connected
5GB EACH for IFR and OFR.
SO we have 3 RAS per server and each server has 6 cores that is SHARED amongst CMS / OS / BOE / RAS..
your thoughts on each points will be awesomely appreciated.
There are CPU and memory requirements for the RAS :
2. RAM :
When a report is viewed and loaded into memory, the report is decompressed to 40 times the report size.
So you need as much memory as :
average report size * 40 * maximum simultaneous processing jobs