Attachment uploads are currently disabled. This is a temporary situation and will resume as normal in the coming days.
UNSOLVED
A
Anil2sharma
36 Posts
0
2716
January 15th, 2015 23:00
VNX SP Utilization threshold values??
Dear Friends,
I need to understand and explain the insights of the VNX storage processor ( SP A and SP B ) utilization to my clients..
Need help on the below questions:-
1) What would the maximum threshold limits of SP A or SP B utilization with out any performance issue ?
2) And for instance the SP A utilization is 45% and SP B utilization is 45 % too. And by chance is SP A goes down.
Sp in that case, does the utilization of SP B goes to 90% OR it can should be less 45 % ( SP B own utilization ) + some minimal X% utilization for the over head gained from SP A.
3) Is there any white paper or video which explains all this concepts and says something about SP utilization behavior.
I am in urgent need of these answer. Guys I appreciate your valuable replies.
35% both SPs, you would have ~70% max if one SP goes down. Safely one SP would manage the load from both SPs. Above 70% you can start getting latency from the component in case. But each environment is different and should be designed accordingly. Also the write cache would be turned off and that would result in direct access to the disks, read/write from that SP.
IaKK83AaCI12227
29 Posts
1544
0
Posted January 16th, 2015 06:00
As far as "the write cache would be turned off and that would result in direct access to the disks, read/write from that SP", this is not the case with all VNX array generations. Back in the CX4/FLARE 28 days, a new advanced cache architecture was introduced. All VNX array generations use this same advanced cache architecture approach.
As far as VNX related documents, there are "Introduction to the EMC VNX Series - A Detailed Review" white paper for both the VNX1 and VNX2 Series arrays available for download.
Good reading!
Thanks Alexander, Miss read it on KB.
- For CX4 and first generation VNX storage systems (Releases 31 & 32), write cache will not be disabled for the following:
- NDU: FLARE Code Upgrade
- Single SP restart
- Single SP physically removed
- SP or I/O module replacement or repair causing SP restart
- SPS Failure (if both SPS's are really faulted, write cache will be disabled - there may be cable issues that mask this).
- Single SP hard fault or transient hardware failure
- Power Supply failure
- For CX4 and first generation VNX storage systems (Releases 31 & 32), write cache will not be disabled for the following:
another counters to keep in mind besides SP utilization:
http://thesanguy.com/2012/09/25/san-performance-metrics/
great reading
brettesinclair
2 Intern
•
715 Posts
1543
0
Posted January 19th, 2015 18:00
Andre Leite wrote:
35% both SPs, you would have ~70% max if one SP goes down. Safely one SP would manage the load from both SPs. Above 70% you can start getting latency from the component in case. But each environment is different and should be designed accordingly. Also the write cache would be turned off and that would result in direct access to the disks, read/write from that SP.
Write Cache is mirrored so wouldn't (ordinarily) be turned off/affected in the event of a single SP failure.
khanz1
2 Intern
•
136 Posts
1544
0
Posted January 16th, 2015 00:00
I recommend taking a look at the Best Practices guide for VNX