Benchmarks - Iometer
So if we can't find an Achilles heel by running simpler benchmarks, how about Iometer?
Simulating a possible real-world scenario, our database workload consists of random 8KB transfers in a 67 per cent read/33 per cent write distribution. Measured in operations per second, performance tails off as the number of users is increased. This is unexpected behaviour - the more you can load an SSD, the faster it should become - but do realise that the MX100 is faster than the M500 in every instance.
The sheer throughput of a datacentre SSD puts consumer drives to shame. Comparing one generation to the next, the MX100 scores 5,780 IOPs to the M500's 4,871 IOPs at a queue depth of 32.
The workstation pattern makes use of 8KB transfers in an 80 per cent read/20 per cent write distribution. Performance should be taken in context here. The MX100's score of 9,412 (QD32) is 35 per cent higher than the M500's, even though the workload is light on writes, where you'd typically expect the 512GB drive to shine.
