Forum Discussion
NimbostratusBest Practice for SAML, F5
I currently have a situation where the team as designed the following for SAML:
Active/Passive If server A goes dead, server B becomes active. When server A comes back online, I now have an active/active situation and things get broken pretty fast. The team is pointing the finger at F5 but this to me sounds like a bad SAML design.
My question is then what would be a best practice architecture if we have the following:
Internet->F5->(Web Tier) Apache RP->F5 (App Tier) App Servers
I've searched for a bit and can't seem to nail down what I'm looking for, with a specific context to F5.
Any thoughts, links, or suggestions is greatly appreciated.
4 Replies
Michael_Koyfma1Cirrus
What exactly is Active/Active? F5s? Or Servers behind F5? Where is SAML authentication happening?
D__Yutzy_151141The way the F5 is configured today is it routes all SAML requests to a single server, if that server becomes unresponsive then it routes all traffic to a second server. This is what I refer to as actrive/passive.
When this event occurs, the second server becomes the primary, active server. If the first server comes back online, I now have a situation where the F5 recognizes both as active and begins to route traffic to "both" servers, it doesn't revert back to one = primary and two = passive.
- D__Yutzy_151141
I'm not at all syaing this is correct, in fact I think it's not which is why I am asking the experts here on best practice and recommendations.
I think that you should probably have both active at all times and configure your pool load balancing to use both based on your requirements. If one node in the pool goes down, traffic will be routed to the remaining active node in the pool. Any connections to the inactive node when it failed will be routed to the active node in which users will have to re-authenticate with the SAML IdP. When the inactive node comes back online, it will be available for new connections. Existing connections to the active node will continue.
