Hi all,
It’s unlikely that our budget will allow us to purchase expensive
Hardware-based Clustering equipment - as recommended by the Shibboleth
team - to achieve IdP Clustering.
Is there a way to achieve High availability, fail over and load
balancing with a software based approach such as with Microsoft NLB (our
IdP is configurated on Windows 2003 server) ?
Would it be possible to use terracotta (or something similar) to
replicate the IdP’s in-memory state between nodes clustered with NLB so
that we would achieve High Availability with this model ?
Another question specifically on the documentation. According to the
section “Configuring IdP Clustering”, in the terracotta model one server
will run in an active mode while the others run in passive mode. But
does this provide Load Balancing ? Isn’t all the traffic being directed
to the Active node ?
Any advice would be much appreciated,
Kind Regards
Karen
|