On 28 Jan 2011, at 20:24, Gary Gatling wrote:
I am in charge of several afs servers in our college. Right now there are 5 afs servers running on 5 SPARC based servers. We are ditching Solaris since it sucks so bad and are going to move to Linux VM's running inside of VMware.
Firstly, I would be cautious about running I/O intensive services like fileservers within a VM. You'll almost certainly get better performance from bare metal, especially if you end up sharing the same physical hardware between multiple fileservers.
I was asked how much disk and memory we would need for the VMs. I was then critized for suggesting that each VM needs 4GB just like the real servers. I did not suggest the 4GB in each server. That was decided by a guy who quit a few years ago.
So, there are two separate considerations here. The first is making sure that the processes on the machine have sufficient memory. The key thing here is tuning your fileserver to suit its workload, and making sure that you have enough callbacks to handle your number of clients. You should be able to take a look at your existing servers, and see how much memory the fileserver processes on them are consuming, and use this as a rough guide. You definitely want to be in a position where there is no way that your fileservers end up swapping.
The second consideration is page cache. Linux uses all of the "spare" memory on a machine as a backing cache for the disk. Depending on the working set of your fileservers, this can have significant performance benefits. Without doing any analysis, it's hard to say how much memory is necessary here, but in general, the more memory you can have as cache, the better.
I don't think that 4G of memory of each of your 5 fileservers sounds unreasonable.
Hope that helps, Simon. _______________________________________________ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
