2011-05-09 09:05 AM
ive been told that our filer might have troubles with misalligned luns which causes high cpu values.
where do i start checking for allignment ?
ive checked a few volumes via reallocation but im not sure what about the result (optimisation 2) or what values should come out of it.
also id like to know if i just have to do it on all normale volumes which have vms on it or how do i combine it with vmdk allignment from the vmkd ? both necessary ?
2011-05-09 12:22 PM
Here is a link to the NetApp Best Practices for File System Alignment in Virtual Environments TR3747.pdf. https://fieldportal.netapp.com/ci_getfile.asp?method=1&uid=5882&docid=30129 It explains how to check in detail (manually or use mbrscan) and you can also look at lun stats or pw_partial_writes in a perfstat with support.
2011-05-09 01:39 PM
For vmware environments OnCommand Insight Balance (IB) (former Akorri BalancePoint) can help you see the partitions offset for vmdks on NetApp arrays (no matter iscsi, fc, or nfs to the esx server). This is also being enhanced in the follow-on 3.6 NetApp to help in other configurations. See attached sample report out IB. In this example you can see it is 3.5 instead of 4KB offset (default for windows 2003 i believe).
2011-05-10 02:53 AM
You probably want to paste a link that is available for non-NetApp and non-partner viewing.
The TR's aren't so hard to find anyway if one just goes to www.netapp.com and searches in the Library tab for Technical Reports and just enters the TR number.
2011-05-10 02:46 PM
here is an example of the pdf report i posted - picks up the vms / vmdk partitions that are having issues ... some great enhancements hit in the next version. Has been very helpful to a bunch of NetApp customers in the past 3 months.