High availability across a WAN link using Heartbeat

Hey guys, im currently doing some research on implementing a linux high availability system across a WAN link and would like help on certain areas. This is a scaled down personal project so I might end up using router simulators like gns3, vyatta etc instead of real hardware. Also it'll involve two webservers on each side on the WAN(actually both server's will be next to each other so it's more like a fake WAN connection).

First, from peoples experience, what are good choices of WAN connections (frame relay, leased lines etc) for such a system since a major concern is latency and since this will across a 'fake WAN' Heartbeat communication has got to be fast.

Second, is it even possible to transmit heartbeats across the WAN. I'm sure latency will be a concern but like I said this is scaled down and is for testing..

From reading linux-ha.org's tutorials and other sources on the Internet, I found that most setups are on same LAN and a serial(RS232) connection is used between the servers/machines for Heartbeat communications. I can barely find good info on setup across a WAN (fake or real).

In precis, I want to be able to achieve failover across a "fake WAN" (if you like). I should add that this will be carried out using Centos5 distro.

Any responses or even pointers will be appreciated.

Similar diagram of what I'm talking about courtesy of dell.com
www.dell.com/content/topics/global.aspx/...s=biz&redirect=1 Please scroll down the page to view the diagram.

Ps. more info about linux HA here: www.linux-ha.org