In our enviroment we have two data center , cross site replication is configured via Brocade 7800 over comcast link at distance over 11 miles.
we have about 7 consistancy groups , including oracle, vmware etc, total data is apx 60+ TB in pair state.
Recently i had to reconfigure Brocades to heigher commit rates, see below.
We have 9 GB bandwith now , which use to be 4 GB before the change for comcast link.
Adaptive Rate limiting is on the switches
Before change: Switches were limited to total 2 GB ( -10% overhead)
portcfg fcipcircuit 16 modify 1 -b 250000 -B 450000
After: Increased limits up to 4 GB max ( -10% ) safety, which is max the switches can due as we have 4 GB uplink from switch to network.
This change was done to all 4 interfaces between DR and Prod, HUR has two Target and two Receiving ports, total of 4 GE ports ( copper) on each end of the site.
portcfg fcipcircuit 16 modify 1 -b 500000 -B 900000
Issues: During the change HUR LDEV were mostly in pair and were not suspended
Once i increased the limits one link at a time, HUR replication Jornal started to fill, replication fell behind and after days of troubleshooting the only thing came out of it was
1. Increase default timeout in horcm file
2. Decreased timeout for ACK to 1000 from 3000.
1. Has any one experianced this sort of issue
2. What do you think caused our problem , currently we have turned it back to what it was , as it appears increasing the speed may have caused duplicate transmit to increase and caused problems
3. With all the troubleshooting, i am not seeying errors/issues on brocade/network or HUR , however, the jornal filled up and replication kept lagging behind, our copy pace on HUR is set to low perposely.
4. What can and should we do to make this go heiger again and what should be the approch.
Is there any addtional debuging we could have turned on for network/hur ?