2

We have critical servers (webservers and databases) that are fully replicated, except for the reverse proxy that we use to hide the internal stuff. This proxy is acting like a router that filters and redirects traffic to the main server and switch for failover if the main one is down.

We want to find an alternative to this proxy because one single entry point is not enough.

Is there any company that has a solid and redundant infrastructure that offers redirection to an IP and allows quick switching to another one?

HopelessN00b
  • 53,385
  • 32
  • 133
  • 208
isoman
  • 203
  • 3
  • 7

4 Answers4

2

You could go with a multi-load balancer solution with a virtual IP (VIP). This would be my suggestion if you're referring to systems at the same location. (you did not specify where you were replicating to)

ewwhite
  • 194,921
  • 91
  • 434
  • 799
2

This is often solved on-the-cheap by having round-robin DNS entries pointing to all of your Reverse-Proxies (plural). You can of course solve this with big dollars (BigIP, F5, etc), but you have to decide what your budget looks like.

If you're not using them already, you should look into Varnish, NGINX, HAProxy, and possibly Apache Traffic Server (which is not the same as the Apache Web Server).

gWaldo
  • 11,887
  • 8
  • 41
  • 68
1

Two things to consider: haproxy, especially since you are using HTTP (and maybe HTTPS) which it is originally designed for. That solves your failover-problem.

The other thing is heartbeatd, which (when installed and configured right) should help you have your one external resource (the incoming IP) accessible through multiple servers (well, one at the time, but if one goes down another can take over, which solves this).

haproxy can be found here: http://haproxy.1wt.eu/

heartbeatd is here: http://www.linux-ha.org/wiki/Heartbeat

flinkflonk
  • 654
  • 4
  • 5
0

'RSF-1' from High-Availability.com fails over the vip associated with the service, and the framework as a whole is pretty flexible

Awalias
  • 154
  • 5