Are there any drawbacks/limitations/gotchas when setting the path selection policy to Round Robin? It seems to me that the Round Robin policy would be the most flexible and efficient since it uses all available paths.
Last week TBL Networks held an #AskHarley event on twitter. This is a pretty neat idea where you submit your virtualization questions to them via twitter/email/facebook and their resident VCDX, Harley Stagner, answers them for you by tweet. I submitted a question about path selection policies which is posted below along with Harley’s answer:
“Ask Harley: Question 1 – What are the reasons / cases for using different VMware NMP PSPs? Why would you not use Round-Robin? Thanks!
Answer:
The VMware NMP PSP that is chosen (either MRU or Fixed) by default will typically be the supported PSP for the storage array.
You should not change to Fixed or MRU from the default.
Changing to Round-Robin is supported on all arrays unless specifically noted by the vendor.
Having said that, different arrays will have different support of the Round-Robin functionality.
More info about path selection policies can be found here: http://bit.ly/ltblbs“
The VMware KB: Multipathing policies in ESX 4 does give a good explanation of the way each of the native mutlipathing path selection policies work but it does not give much insight into the best practices for using them or if there are any drawbacks to using the Round Robin policy with a supported array.
If the storage array supports it and you are not using Microsoft Cluster Service are there any other reasons why you would not use the Round Robin Path Selection Policy? Is there a best practice?
