We are running a LAMP platform which has the following hardware:
-Two Dell R200's running Red Hat (one separate website on each) serving as web servers
-Cisco Smart switch
-One Dell 2950 w/32 gb RAM, 2Xquad core XEON 2.3 ghtz CPU's and 8 x 600gb SAS 10k RPB drives (two RAID groups 1- RAID1 for Solaris OS and 2- 5+1 RAID5 w/hot spare
Things seem to be going well until about a month ago, when the performance of the websites would take huge latency hits whenever we ran a database job in the background. Bear in mind that we have been doing this in production since early November, 2013.
Now, however, whenever we run one of these jobs (e.g., updating tables, backing up database, etc.) the response rate for the server tanks severely.
We've absolutely ruled out the issues being related to the network or the Web servers (after many hours of diagnostics).
We carefully checked out all aspects of the Dell 2950. We've run Dell diagnostics against the entire system (as well as Solaris CPU, memory test commands) and it reports that all of the CPU threads are healthy, the RAM is 100% operational in the disk drives in the raid groups are fine.
We decided that the volumes on the server are exceeding some performance threshold and we think that we're going to drop two gigabit network cards (Intel E1G42ET) into the 2950 and move everything into a Dell 3200 ISCSI San.
We could then re-create the luns across 12 drives (and probably continue to use the 2950's first raid one group for the OS).
My question is: will this alleviate these latency issues that have popped up over the last month? What might be a better approach?