You are right. It is no that simple.
Your FC drives are able to provide on average that backend performance.. Given that each IO is on average 1/2 a rotation away on the platter. If you have sequential you get more IOPS, if you need to rotate more than 1/2 platter on average it will be less....
But this is only a small fraction of the total picture... Each R5 write IO will produce 4 backend IO... On R6 it is somewhere between 6 and 7.
On the positive side you have cache, where a cache hit is a backend IO you don't have to do.... And with SSDs, I'm assuming AO which bring the effectiveness of that into the picture.
So my best suggestion is either to look at what you're doing today (frontend IOPS), look at %busy on backend (disks) and assume that the new dat is like the current data and do those numbers.
_________________ The views and opinions expressed are my own and do not necessarily reflect those of my current or previous employers.
|