We have 3 AS/400 systems PR, HA (Both are on same physical location) & DR (situated in different physical location around 1000 KM away from PR system).
All the three systems are in sync using MIMIX Global. We had 100 mbps WAN link configured for PR to DR replication. PR & HA replication is on internal network.
We are using remote journals for MIMIX.
We have started facing RJ lag during night time when there are huge journal entries getting created due to End of the Day jobs. This Remote Journal lag goes beyond 30 min. & 15 million transactions on few days.
Hence management has decided to increase the network bandwidth & not we are on 200 mbps WAN link. Still no improvement in network performance.
We are not having any replication problem between PR & HA as these systems are in same data center connected on local LAN.
As confirmed by our network team iSeries system are not utilizing the full bandwidth and bandwidth utilization is only 40%.
We are not having any RJ lag during normal business hours.
We have checked Retransmissions but they are not much. MTU value also set to *IFC.
The questions are,
1) Will increasing/decreasing the TCP Buffer Size on source & target systems help us to improve network performance?
& 2) How to identify the exact TCP Buffer size required for best network performance?