Question

I am developing a tcp/ip server whose purpose is receive packets from client, parse them, do some computation(on data arriving in packet) and store it in database. Till now, everything was being done by single server application written using twisted python. Now I am across RabbitMQ so my question is, if it is possible and if it will lead to better performance if my twisted server application just receives the packets from clients and pass it another c++ application using RabbitMQ. The c++ application will in turn parse packets, do computation on it etc.. Everything will be done on single server.

Était-ce utile?

La solution

If your server is does not receive packets often, it will not improve much - only gain some tiny overhead on inter server communication. Still it is a very good design idea, because it scales well and once you finally get many packets you will just add an instance of data processing server.

Licencié sous: CC-BY-SA avec attribution
Non affilié à StackOverflow
scroll top