Notifications
Clear all
Topic starter
We have 2 remote sites that are connected back to our main site (where database & Jboss are located) by 2 Gbps fiber ELan. Our exam room PCs have CPS installed locally and when on Windows 7 thick clients, everything runs fine. When we roll out Windows 10 thick clients with CPS installed locally, CPS randomly freezes and is borderline unusable.
Wondering if anyone else has encountered this?
Posted : June 19, 2020 3:04 am
Are you routing (and potentially encrypting) traffic between the remote sites and main site, or are you switching the traffic between sites?
Either way, I would probably start with analyzing a packet capture of a sample of the network traffic on both Windows 7 and Windows 10. I'd set up a test scenario such as open registration for PID 12345, then open the chart for PID 12345 so that the captures should be very similar.
The Windows 7 sample would be a reference to see what it looks like when things work as expected. You may not need the Windows 7 sample, but since you're in a position to get it easily, I would.
Then analyze the packet capture from Windows 10. My guess is you'll end up seeing varying degrees of client retransmits after timeouts, but there are many other possibilities also. It may also help to get a packet capture on the database server at the same time.
There are several possibilities. A problem with path MTU discovery, firewall rules, or encryption and/or 802.1q frame overhead on the ELAN link.
-dp
Posted : June 19, 2020 3:41 am
Topic starter
Thanks Dave... So in your experience, this is an acceptable workflow and and we shouldn't be seeing these freezes, correct? We've been digging into our network a lot lately and will also be involving the ELAN provider next week.
Posted : June 19, 2020 4:03 am
I don't see a problem with the workflow as long as latency is reasonable.
I remember back in 2006 testing a fat client over a VPN with a latency of maybe 20ms at the time. It was terrible. We use terminal servers for 100% of our users because the client/server communication are so latency-sensitive.
The client mostly does SQL queries to the database server, and it isn't efficient about it. The queries are broad and the results are broad, resulting in the back and forth being quite chatty. IIRC, a bit of the data now might flow to JBoss now, but it isn't much.
I definitely don't think you should be experiencing the issues you describe.
-dp
[edit: altered confusing statement]
Posted : June 19, 2020 4:51 am
Make sure network drivers are up to date on the Windows 10 boxes. Might exclude the CPS directory from any antivirus/malware scanning. Might even disable TCP/IP 6 on those machines if you don't use it.
Posted : June 19, 2020 12:35 pm
