We're migrating from TMG 2010 to UTM 9. Everything appear to be OK during testing, but once we went into production and the traffic increased we realized that all the traffic from our UTM through the Barracuda Networks load balancers were going to only one of our four web servers. I tried disabling connection pooling and http keep alive, but we get the same result. We eventually had to fail back.
The TMG passes the original client IP to the load balancers, whereas UTM doesn't provide that option. In theory Barracuda should not be using IP's as the sole determining factor for distributing traffic. I am checking on that also, but wanted to see if anyone is aware of any other configuration options that may mitigate this issue from the UTM side.
This thread was automatically locked due to age.
