#Latency in on Reading / Retrieving data from StorageGrid with FabricPool

1 messages · Page 1 of 1 (latest)

smoky jackal
#

We are experiencing a slowness on a Oracle RMAN restoring process. The backup data on SSD aggregate with FabricPool plicy = auto, 7 days minimum. The data required to be restore are 20 days old, so, the data likely are already in SG. Usually it only took about 5 hours. But, now it is taking more than 20 hours and yet only 50% is done .

To me, the slowness is not just due to reading data from Object Storage. There might be something else wrong. Do you agree with this assessment?

What is usual reading latency from SG in ms?
What can we do to determine if the slowness is really due to read from SG?

hot hatch
#

on the GRID Site you are able to create a Traffic Classification Monitor (without Limits) on the bucket/Tenant the Data are tiered out.
maybe that helps to identify how Big / Small den Objects you are reading from the Fabric Pool. And see if the Bandwith is caps ont he WAN/LAN or something..

if you have a high latency the Ontap will report this in the event log. (i think 1000ms is the Threshold)
(so thats math then... small object reads + Bandwith = max Throughtput)

from the Ontap side, you can go to ne Node Shell and in "priv set diag" you see the Cloud thoughput with "sysstat -b" in realtime.

maybe it helps...

distant flower