Jasmin SMSC throughput

One of our recent project we started using Jasmin SMSC and run into low throughput issues details shared with Jookies are here

It turns out that the Interceptors over TCP have long lived TCP connections and cannot be scaled so we came up with our own solution of moving interceptor logic to a separate microservice.

We performance tested the service independently and once we were sure there we no bottlenecks we modified the Jasmin SMSC source to use a HTTP interceptor instead of a stock TCP based interceptor.

We have dockerized Jasmin SMSC as well as the Interceptors microservice and thus allowing us to scale horizontally.

So it looked something like this

There are a few challenges here you need to have enough binds allowed by your providers so the SMPP sessions downstream are allowed on all Jasmin SMSC dockers.

If this is not possible you will have to separate the receiving Jasmin instances and downstream sender Jasmin instances (haven’t tested this approach yet still theory)