(Migrated) Network addresses

(This message has been automatically imported from the retired mailing list)

I can see a potential use case where you have servers on remote sites
that can only access the Redis instance and the load balancer. It
might not be desirable to open up the firewall between all of the

In that scenario, the main site servers would be load balanced. All
servers would call shared services via outgoing channels back to the
load balancer rather than directly invoking them to ensure that they
ran on the main site cluster.

Clients on the remote sites would call services directly on the local
server, rather than via the load balancer, or would call a local load
balancer if you wanted resilience on each site.

It’s the kind of setup you often see with monitoring systems that
deploy remote pollers to different sites (or even just different
VLANs). The pollers can see the master server but not each other.

Of course, an alternative that may be even cleaner (but require more
effort to maintain) would be separate Zato instances on each site,
where the instances can pull service configuration from the master