RTC component load is too slow in certain client machines
We tried load component from a stream and it take a long time as the data transfer rate is in KBps. Hence timeout issue occured and fails to load.This issue happens in certain remote machine of a same IP segement.
But the same task is working fine with data transfer rate in MBps. When we analyzed the network performance in the machine (bad transfer rate) it is been seen there is no any network latency.
Actually not sure what is the exact problem is. Can you please provide suggestion how this to be troubleshooted.
2 answers
RTC also introduced the Metronome Tool for profiling network issues with a RTC repository connection.
In Eclipse this should be at the bottom of the screen. If you select it you can try a connection speed test on the 'Connection' tab, it also gives you the cache size.
The Services tab will give you timing metrics, you may want to spot-check a few compared to a "good" machine to make sure latency is not your biggest issue.
In Eclipse this should be at the bottom of the screen. If you select it you can try a connection speed test on the 'Connection' tab, it also gives you the cache size.
The Services tab will give you timing metrics, you may want to spot-check a few compared to a "good" machine to make sure latency is not your biggest issue.
Do not forget the possible effects of a virus scanner either. Use a performace tool ( e.g. windows task manager, 'top' command ) to see if there are processes taking a fair percentage of CPU and can be associated with an executable in the virus scanner's suite of executables.
Another factor could be a too low value of MTU on the client TCP stack. Use the tracert (traceroute - unix o/s) from the RTC host to the client machine. If you see any "fragmentation required" that will be a throughput killer.
Another factor could be a too low value of MTU on the client TCP stack. Use the tracert (traceroute - unix o/s) from the RTC host to the client machine. If you see any "fragmentation required" that will be a throughput killer.
Comments
Krzysztof Kaźmierczyk
Aug 07 '14, 6:03 a.m.Hi Chandrasekar,
I would recommend you to open new PMR to make a complex analyzis here. What I already know is that sometimes loading component after long inactivity causes performance degradation. It is caused by the fast that RTC server uses internal cache to store data.
You can also try slightly increasing -Xmn value in your Virtual Machine settings in WAS profile or in server.startup if you use Tomcat.