Open hty0506 opened 4 years ago
The latency monitored by Jaeger is inclusive, which contains the processing time of the service and the latency of down-streaming services, including processing and network. There is no trivial solution to separate processing time and network latency without instrumenting the network stack into the Linux kernel.
@gy1005, so , how did you guys extract the communication time or the network delay and statistics? Any tips? Even the paper measures Computation:communication ratio for different services. Any help on how it was done as none of the scripts in socialnetwork correspond to recording network/communication time.
Communication method, how to obtain the communication trace between microservices,?how to obtain the RPC processing time?