i’m pretty struggling with a problem which is impossible to see from a “down”-layer view. While PRTG/Smokeping and so on only see how the average RTT goes from, for example, average 5ms to 20ms 20ms 20ms - we have customers complaining about the interruptions and cuts on an upper-layer protocol like tools in DC to branch-office. What you can see is retrans. on TCP session but windows size is always good and whis 32-64bit msgs. Maybe you can advice?
Thank you (ps access is always clean and we do not see any capacity problems).