Analysis
On an example of a breaking application using UDP, such as RDP on Windows 10, use Wireshark to capture some network traffic.
After some time, when the issue happens, filter for the UDP destination port (in this case, 3389), then select the first UDP stream that corresponds to the application.
On RDP, consider the packet timestamp value when the issue occurs. You will notice that, after a while, the UDP communication happens every 15 seconds.
The VP network engine has the default UDP timeout value set to 10 seconds. In such a scenario as depicted above, the connection will get dropped and the UDP stream will have to restart.
The above behavior is the same also for other UDP based applications.
Solution
There are two solutions:
- Increase the UDP Timeout value (and optionally also max UDP Connections); or
- Create a bypass rule for the UDP services
How to increase UDP Timeout
A suggested value to avoid those issues is:
UDP Timeout: 45 Sec
It is also suggested to increase Maximum UDP Connection.
Maximum UDP Connection: This is optional to increase
- If you already have a custom value, it is suggested to add at least 30% on top of the current value shown in the VP manager console
- If you have a standard value and VP endpoints are expected to have high traffic, you could set it to 50,000
To change these settings for all computers, you can apply the changes on the policy level:
- Go to Policy > Settings > Network Engine > Advanced Network Engine.
- Modify the values highlighted in the image below: