When there is spillover for a combination that should be fully optimized (that is, a session that should be optimized is not optimized because the WAN-optimization device cannot take another session at that moment), SPAN measurements for the spillover sessions go to the [Server] segment.
No measurements for the spillover sessions affect the [Client] or [WAN] segments. Server Response Time [Server] is accurate, even though it contains measurements for spillover and optimized sessions. All metrics on the Optimization page are accurate. Network Round Trip Time [Server] shows an increase because it no longer gets 100% local ACKs. The spillover measurements show actual Network Round Trip Time out to the client. RetransDelay[Server] and PacketLossPct[Server] might also show increases.
In a properly sized WAN-optimization deployment, there should be little to no spillover. If you find a consistent difference between off hours and high-load hours for Network Round Trip Time [Server], you might have spillover. The solution is to increase the capacity of the WAN-optimization deployment. If all networks show the increase, the data center WAN-optimization device might lack adequate capacity. If only a few networks show the increase, those branch WAN-optimization devices might lack adequate capacity.
Another clue to spillover is when more sessions are reported on the [Server] segment than on the [Client] segment. This happens when non-optimized sessions for a combination that has at least some optimization are reported in the [Server] segment rather than in the parent application where they belong.
|
Copyright © 2015 CA Technologies.
All rights reserved.
|
|