I see a following remark on the VNX side. Are there any equivalent best practices around VMAX pools? Or refer the best document you know on this topic.
"You have auto tiering on. Fast_xx_pool is currently set up as 100% used on the top tier. Based on your code level you should be striving to get to 95% used at the highest. 90% is more ideal. The luns need that space to move around."
FAST movements history is not for FAST VP, It is only for FAST DP. (Check the note on page 224 from SE Array controls guide v7.6). You cannot have a full history of extent groups movements for FAST VP (at least not something that customers can view). The only customer viewable option is exactly what my master @paul martin mentioned, use FAST VP ingress/egress tracks movements graph charts from Unisphere for VMAX. Unfortunately, there is no way to track FAST VP movements using SE.
The current configuration you have from symfast list -vp -state output is 100% normal and movements are taking place between your defined tiers.
I believe the configuration of FAST related setting in VMAX should depend on your real environments and workload. If you're looking for the best practices documents, I think the following documents are good to go:
I have reviwed the docs, and one thing still not able to track. I need to review the logs to make sure the the FAST VP operations are happening as scheduled. ( tried symauit already with fast and optimizer application/class). I can see the data movement happening, but i need the logs to confirm.
if you want to check FAST VP movements you can check this at any time using Unisphere for VMAX with performance, drill down in the realtime or diagnostics and check the metrics for the pools, the counters you are concerned with are
ingress tracks - tracks moved into a pool by FAST VP
egress tracks - tracks moved out of a pool by FAST VP
I have tried the plan and history options ( symfast) already and they always return me "no data movement" . I dont see more on the SE guide on tracking.
plan option talks about disk group provisining( so am i dealing at different level , VP)
But i can see the data movement is happening ( see the current state).
This is a very good question, The short answer is No, You do not need to disable FAST VP however, The outage/replacement of SP will impact the FAST controller resides on the SP in 2 ways:
1- Allocation-Compliance algorithm will be completely disabled as FAST controller is solely responsible for the out of compliance movements.
2- Intelligent-tiering algorithm will not be under big impact as Enginity will use last good known thresholds for 4 hours and then backup thresholds for 4 days. After the 4 days. You can say FAST VP is completely disabled (no more Intelligent tiering or allocation compliance movements will take place).
For more details, kindly read page 34 of FAST VP theory of operation Whitepaper:
SP replacements should be immediate as many other SYMMETRIX vital functionalists will not work like configuration changes but in your case (planned outage for replacement) the impact will be minimal.
M_Salem
213 Posts
1628
2
Posted September 1st, 2014 07:00
Hi SKT,
FAST movements history is not for FAST VP, It is only for FAST DP. (Check the note on page 224 from SE Array controls guide v7.6). You cannot have a full history of extent groups movements for FAST VP (at least not something that customers can view). The only customer viewable option is exactly what my master @paul martin mentioned, use FAST VP ingress/egress tracks movements graph charts from Unisphere for VMAX. Unfortunately, there is no way to track FAST VP movements using SE.
The current configuration you have from symfast list -vp -state output is 100% normal and movements are taking place between your defined tiers.
Hope this helps
Mohammed Salem