HPE Storage Users Group

Hi,

I am trying to come up with a sensible formula in setting my max IOPS on my all_others group within QOS policy.

Say I have 100 15k FC Disks. I can calculate the maximum iops they would produce in best case and assume 100 Disks * 300 IOPS would give me 30,000 iops. Not taking caching into consideration just yet, does anyone know how many back-end iops are produced for every single front-end iop for RAID1 and RAID5?

I know the write penalty is not the same as traditional raid which is why I am posting this question.

Thanks!

This is subjective in that it depends what you're trying to achieve. It's easy to over think and more importantly over use the QOS rules. You can do all these calculations front end vs backend but read write ratios, block sizes, raid levels cache hits etc will make this a moving target.

My understanding is that all others is a catch all for where you want to be able to provide some level of QOS control / throttling but you don't want to actively manage each and every application added and its associated QOS rules.

As such I'd look at setting the max value for all others to either whatever you think those apps will require (not easy to know or keep on top of) or more likely something bigger than the box can actually supply whilst also setting the all others group to a lower priority than your other QOS rules. Each of which should also have min goal. That way the all others group will be throttled first to meet the min goal of the high priority QOS rules and so in theory, should get whatever's left over after your other higher priority QOS goals are met.