We are using eMatrix (Engineering Cental version 9.5) PDM system.
Usability is not very satisfying because of long responce time in the system.
We are using the system in europe, asia and america.
System vendor is saying all the time that the problem is in your network. The ping round trip time is in europe less that 100ms and in america and asia something like 330ms.
When we monitor the server computer CPU load and memory usage it seems like everything is ok in there.
We are using the ematrix trough web clients and on server side we have dedicated oracle server and application server (BEA weblogic).
My question really is that where is the bottle neck?
A) Is it between the client and application server or
B) Is it between the application server and Oracle server.
Many people say that the eMatrix problem is that it is doing many round trips. Is that meaning that round trips are done A choice or B choice?
How can the number of round trips easily monitored?
Check your server-to-server connectivity using the ping command - how many milliseconds does it take. Similarly, use the tracert command to get a detailed list of network hops and how long each hop takes.
AFTER you have done the above two checks to get some baseline numbers, check your connectivity from App Server to DB Server using the tnsping SID TRIES command, where SID is the name of the database whose listener you want to hit, and TRIES is an interger number of times to repeat the test.
Keep in mind that many products were not designed for wide area networks like transatlantic fiber or satcom links. Ask the vendor if they specifically designed it for transatlantic WAN, or if it was designed for use in a campus WAN. Their answer might be "we didn't do either, and it should work just as well over 5,000 miles as it does 50 miles." Yah, right! They might also be sending too much overhead info, so at each multiplexing poing or other network bottleneck you have to wait longer. A few extra millisecond times several dozens hops = a lot of extra milliseconds.