We have a DVA Citrix ICA SS. We have many operations coming through from different locations in Australia.
We have a lot of operations coming through that are extremely slow (~10s). Most of these operations are CTXTW (Thinwire) - Remote Windows Screen update Data.
Interestingly, the Operation time with breakdown shows ALL of the operation time under Network time and zero for server and redirect time. The traffic is not unidirectional.
Any thoughts on this??
Please note that Citrx ICA operations are unidirectional by their nature. Although ICA uses TCP as the transport protocol, there is no logical correlation between client->server and server->client streams. Client sends information on keystrokes and mouse movements, server sends commands to update the screen. Both sides buffer and bulk-send sometimes, depending on the sophisticated algorithms that ICA uses for optimization of the user experience.
DC RUM measures "command delivery time" for ICA, that is the time it took server to send to client a declared number of bytes commanding the client to update a portion of the screen. Citrix server, when it starts to send a screen update, declares how many bytes would it be sending - we count it as the transmission progresses and this way we know what was the command delivery time. Note here - there is no server time component in it, as there was no client request for update - server may start update at any moment in time, even with no user action at all (think of a web page that has animated content witch refreshes itself - from the Citrix display perspective this is 100% unilateral server directive to paint apportion of the screen).
Command Delivery Time values greatly depend on the type and characteristics of the application delivered over Citrix. Therefore looking at absolute values of this metric may be puzzling. A much better metric is the Realized Bandwidth, which is representing effective throughput that the client experienced when receiving the screen updates. This metric corresponds to Citrix's comments on best practices for network throughput required by ICA. Realized Bandwidth is the primary metric on Citrix Landing Page reports.
In general, the Operations report accessible as a drill down from the Software Services and Servers is not the right report to analyze Citrix traffic. There is a separate Citrix Landing Page report (and its linked reports) that should be used to analyze Citrix performance. You wouldn't find "operation" term on those reports, so my guess is that if you mention Operation in context of Citrix, you are not looking that the reports you should be looking at (i.e. the Citrix Landing Page).