> 1. What would be an appropriate value for our application for MAXCPU when running on a quad processor machine. The application consists of the MaxDB along with 4 windows services and an instance of tomcat. Can we use 4 for MAXCPU or does it always need to be something less than the total number of CPUs on the machine (less than 4 here?). Please advise.
Of course you may set MAXCPU to 4 - but you risk that the four Usertask UKTs will use up all CPU resources on your system then.
So it is usually better to be rather defensive with this. If you're unsure about the CPU needs of your other services better try out and check starting with MAXCPU 2.
> 2. When we're doing query execution time analysis we need to be able to clear data cache in between query runs. We achieve this so far by simply bringing the database offline and then online again. The problem with this approach is its time consuming and if we do 100 query runs a lot of time is spent in waiting for the database go from online to offline to online again. Please let us know if there's an alternative faster way to achieve the same.
Well, x_cons pagecache_release should do that for you.
But I never was able to really see the effect of it (e.g. reading the pages in again).
If I'm not mistaken in this point - which is pretty easy because the command is rarely used and sparsely documented - it just moves the pages to the lower end of the lru list and thereby making them candidates for getting moved out of cache but not actually doing it.
However why don't you just use the resource monitor to figure out how many pages are touched by your statements?
Then you don't have to bounce your instance.
regards,
Lars
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.