assuming that everything else in your test is the same and you arent seeing other side effects keep in mind you are testing two very different configs
when you use RDM your client I/O gets sent unchanged to the Unity
when you use a vmdk in a ESX datastore you are going through the vmfs file system on the ESX server. Not only does that have some overhead it will also change the I/O type and block size
Take a look at the performance stats on Unity - you should see differences there
Its not really "fair" to compare these - vmfs has quite some benefits and functionality that arent available with RDM
Rainer_EMC
4 Operator
•
8.6K Posts
0
March 4th, 2019 06:00
assuming that everything else in your test is the same and you arent seeing other side effects
keep in mind you are testing two very different configs
when you use RDM your client I/O gets sent unchanged to the Unity
when you use a vmdk in a ESX datastore you are going through the vmfs file system on the ESX server.
Not only does that have some overhead it will also change the I/O type and block size
Take a look at the performance stats on Unity - you should see differences there
Its not really "fair" to compare these - vmfs has quite some benefits and functionality that arent available with RDM