QuickFIX/J - failover strategy

Question

I would like to ask about a couple of failover strategies for QuickFIX/J and Spring Boot QuickFix starter

For example if I have a FIX engine server and receiving a lot of FIX messages during all day and suddenly the service becomes unavailable.

1. What starts to happen when the service goes up again? Where will it start to read the new FIX messages again?
2. What will happen when the service starts to have a heavy load and kubernetes starts putting a second instance? Is there any way to keep data consistency between two microservices so that they do not process the same message twice?
3. How to deal with multiple sessions on multiple microservices and scaling at the same time

Thanks for response, I'm just starting with this library

Answer 1

1. The FIX engine will synchronise the messages based on the last message's sequence number that it has received. You can read about the basics here: FIX message recovery
   Since you are new to the FIX protocol that whole page might be a good starting point to make yourself acquainted with the protocol. Of course the FIX engine will do the session-level related stuff on its own but it's always good to know the basics.

2. I don't really have any in-depth knowledge of Kubernetes but the important thing here is that a FIX session is a point-to-point connection. That means for the very same session (identified by a SessionID which usually is composed of BeginString (eg FIX.4.4 ), SenderCompID , TargetCompID ) you will only have one Initiator (ie client) and one Acceptor (ie server).
   So spinning up a second instance of a service that will connect to the same FIX session should be avoided. This would probably work if you had several sessions distributed over several instances.

3. Don't really know what you mean by this, sorry.