The SR already opened and below are the recommendation

• 1-      Reducing read size IO from the application where back up is being done as it’ s the main factor aligned with high response time seen on LUNs side & triggering its peak values.
• 2-      Since SAS drives are overloaded during backup time and Extreme performance tier is almost full , we need to add extra flash drives to the pool  and check if our main issue is enhanced , if not , we will need to reduce read size as explained previously.

wi-tribe, at first glance from what you are describing, it seems like the array may not have been sized/configured correctly for your environment's needs.

Having said that, this is not the place for this kind of troubleshooting.

Please open a Service Request for the perf team to take a look.

We would require:

- a very clear description of the issue

- all the time stamps of when the issue was observed (including timezone of the observation so we can match to our logs)

- Unity data collects (https://support.emc.com/kb/483583)

- UPA files (Unity Perf Archives) covering the times in question (https://support.emc.com/kb/491175)

Once we have all of this, we will help you narrow down the cause.

Thanks

Andre @ Dell EMC

Yes, so the issue seems fairly obvious. There is very little we can do if the drives are overloaded.

My initial suspicion was correct where I said: "wi-tribe, at first glance from what you are describing, it seems like the array may not have been sized/configured correctly for your environment's needs."

You need to reach out to you sales rep or implementation partner, or account manager to see if a new full environment assessment can be done, and then design the system based on your requirements.

How will you manage to reduce the read IO size?  When the read is sequential especially during the backup the IO size is going to be higher.

If its block you could also look at QoS

wi-tribe - what is the SR number?

glen

Below are the SR number

SR# 10276103

SR# 91383920

SR# 09999650

SR# 11014415

SR# 10199664

to reduce the size i think we can lower the number of streams.