As per my last entry I've been coming up with a method of visualizing the raw utilization of our storage filers.
The numbers them selves say a whole lot.. but I wont (can't) go into that here other than the value I see is very scary.
I'll quickly go over the algorithm I used to arrive at this number though just to make sure I'm getting what I think I'm getting
First lets define the raw utilization notion. This in my understanding would
be the overall used space divided by the complete space passable (irregardless
of raid overhead, spares etc). So this would look something like:
%utilization = ((used space) / SUM(all-disks-raw-space))*100
Here is the sudo code I came up with to collect this info via SNMP
raw=0
used=0
#gets raw disk info
snmp get raidPTable
foreach @disk
add value to raw
#gets raw spare info
snmp get spareTable
foreach @spare
add value to raw
#gets usage info
snmp get dfTable
foreach @aggr
add value to used
return(raw,used)
Now we just loop each filer through this routine adding up the raw and used return values into a more global space and perform a final calculation to arrive at the total utilization percent.
This added about 23 seconds on to the run time of this filer monitor script so I think I need to flag this option or run this out of band.
Posted by Kevin Foote in work on August 03, 2009
