I just want to double check my findings here, hope someone out there is able to weight in on some of this.
Let's say I have a SSD storage array capable of delivering 25,200 IOPS and it is connected up to my server using 8Gb/s FC link, with an average IO size of 75kb.
The 8Gb/s FC link after 8b/10b encoding overhead gives you 6.8Gb/s usable bandwidth, or in storage terms 870MB/s and 500,000 IOPS theoretically (I am going off by various online blogs I've read).
Using the MBps = (IOPS * KB per IO) /1024 formula with my numbers above, if I was to run a loadgen on the server to max out the 25k IOPS at 75k avg size, this would need 1,845MBps of bandwidth.
The 8Gb/s FC only provides 870MB/s, so my bottle neck in this case is the 8Gb/s FC link?
Any storage architects out there able to help me out?