IO Question

• r5d4

  SSCrazy

  Points: 2519

  More actions

  December 8, 2016 at 7:15 am

  #332658

  Hi there,

  I'm attaching 2 diskspd outputs, from testing a sql data drive with 4k and 64k block sizes.

  As you can see the performance of the latter is fantastic, the former not so good.

  (The drive itself is formatted to a 64k cluster size, if that even still counts for anything nowadays)

  This is a SQL 2016 on Windows 2012 Virtual Server.

  I have NO access to vmware itself, so I need to know -

  a) if there is anything I can do (whether it be tests or configuration)

  b) questions / lines of enquiry for the vmware admin.

  I used diskspd as I was seeing suspended writes in activity monitor and Paul Randal's script for IO latency gives me these figures for the drive -

  Read Latency - 48

  Write Latency - 26

  Overall Latency - 36

  Avg Bytes/Read- 112519

  Avg Bytes/Write- 74563

  Avg Bytes/Transfer -91909

  Thanks

  Rich