We are looking at speeding up our DataStage jobs with the use of SSD technology (enterprise SSD).
Our main bottleneck is the Disk I/O, so the idea is to try and decrease this bottleneck. We would mainly use the SSD for Scratch/Temp/Sorting.
Has anyone implemented SSD's with Information Server ?
Did you face any challenges ?
Did the SSD make much difference ?
Thanks in advance.
Notice: We have upgraded developerWorks Community to the latest version of IBM Connections. For more information, read our upgrade FAQ.
JohnMcKeever 2700050KAF1 Post
Re: DataStage and SSD2012-03-13T05:59:10ZThis is the accepted answer. This is the accepted answer.Hello KJ75 -
We're using DataStage installed entirely on SSD, but have not 'migrated' from a hard disk so am unable to provide a direct comparison.
There were no notable challenges .. it presents to the operating system the same logical volume as a spinning disk.
If disk I/O is your bottleneck then there are a number of approaches you can take to optimise your hard disk usage before you splash out on SSDs.
Without knowing your configuration it's hard to be more specific.
- Have you configured multiple (virtual) disk adapters?
- Are you using a dedicated connection between your server and disk array (rather than going through a shared switch?)
- Are you distributing node data (segment files) evenly across physical disks to achieve maximum concurrency and minimum contention?
- Are your jobs optimised to make maximum and even use of parallelism?
- etc. etc.