Forum Discussion
Load Balancer Session Persistence and Existing Connections Clear Information
Hi,
Going on Azure documentation on session persistence for Load balancers and reading the following on client IP and protocol:
1) It is mentioned in this article that Client IP and protocol settings is used to solve an incompatibility issue between Azure Load balancer and the Remote Desktop Gateway. In the scenario mentioned that TCP and UDP are used with client ip address and protocol. If I understand correctly this setting should send UDP from same client to a different load balancer. So it does not solve this problem.
2) The behavior of existing connections through a load balancer. As per my understanding TCP connections will not be rebalanced to a new backend instance after a failure while UDP will actually be sent to another VM as it is connectionless. is this somehow related to the persistence settings ?
Does Azure have any workaround for this behavior. For example GCP has this feature of rebalancing existing connections.
Thanks.
1 Reply
Azure does not currently offer native TCP connection rebalancing like GCP Load Balancer does. However, here are some options:
1. Use Azure Application Gateway or Azure Front Door
• These offer layer 7 routing, health probes, and better failover handling.
• Application Gateway supports connection draining and session affinity via cookies.
2. Implement retry logic in clients
• For TCP-based apps, ensure clients can detect dropped connections and retry.
3. Use HA architecture with faster health probes
• Reduce probe intervals and thresholds to detect failures faster.
• Combine with autoscaling and zone redundancy.
4. Consider NVA-based load balancing
• Deploy a third-party network virtual appliance (e.g., F5, Barracuda) that supports session-aware failover.
