|
On Fri, Sep 15, 2000 at 08:12:21AM -0400, Ted Pavlic wrote:
> I haven't been keeping up on this particular thread, so I don't know if this
> has already been brought up, but rather than waiting for BGP to timeout,
> I've built a slightly different setup that still uses routing to route the
> IPs directly to the LinuxDirector.
>
> Both LinuxDirectors have their own unique IPs... for example...
>
> 192.168.16.1
> 192.168.16.2
>
> But all the VIPs are actually routed to a third address:
>
> 192.168.16.3
>
> And the active LinuxDirector simply brings up that IP as an alias. I have a
> program (like heartbeat, but I wrote ages ago called etherbeat) which runs
> on both directors and elects a master between them. Once that master is
> elected, the slave brings down its alias, and the master brings up its
> alias. The router's cache is cleared and traffic is routed through the
> master director.
>
> When the master director dies, after a timeout easily tweakable by me in
> etherbeat, the slave is elected master, brings up the alias, and routing
> goes on as it was.
>
> Now I've only *THOUGHT* about using BGP or RIP to update the routing table
> automatically, but this system works great for me. I actually got the idea
> from a similar implementation in my telephone switch.
>
> I don't know if this is better or worse... simpler or more difficult... but
> I know it works as a routing solution -- so I offer it.
What you of described is more or less the current state of play for doing
HA (and Load Balancing) under Linux. Certainly Ultra Monkey makes
references to doing what you describe with heartbeat.
To my mind the idea of using a routing protocol would be best if a routing
protocol is already used on the network. After all if you are using fwmark
to manage CIDR networks that are routed to a Linux Director managing the
path that this traffic takes really is a routing problem and it would make
sense to make use of any existing routing infrastructure. Though would be
thinking more in terms of an IGP such as OSPF rather than an EGP such as
BGP, though this is really an implementation issue.
The question that I have is - and this shows where my knowledge of routing
protocols runs out - what is the relative cost of waiting for various
routing protocols to converts as opposed to using a fail-over mechanism such
as heartbeat or your etherbeat?
--
Horms
|
