4
votes

Our application consists of a stateless web API making concurrent call to a stateful service using service remoting. Through load-testing, we noticed that there is delay caused during inter-service communication (between calling the service and the service itself starting to process the request). This occurs when we increase the load to approximately 1000 concurrent calls per second.

We tried this solution but our timing didnt improve. We then tried to replace our communication stack by WCF with tcp binding but that didnt improve it either.

For inter-service communication in Service Fabric, which communication stack is the most efficient? and any idea on what's causing the bottleneck?

1
Your hardware perhaps?Mardoxx
A cluster of 5 nodes, each a Standard_D2_v2 Azure VM with 2 cores, 7 GiB Memory,100 GiB Local SSDzrvr
Have you looked at the Performance and Scalability samples? Try running that in your cluster to see if your hardware is the bottleneck? azure.microsoft.com/en-us/resources/samples/…Dismissile

1 Answers

0
votes

Try Kestral, it's the fastest that I've found so far.