|
> 1. Co-host or host a server at a multi-homed (no single route failure
> will render it unreachable) facility with redundancy ala heartbeat
> (www).
> 2. the servers which will actually do the work are hosted elsewhere
> at multiple cheaper facilities perhaps, without the required route
> redundancy (www1, www2).
> 3. mon on www monitors the availability of www1 and www2.
Due to the nature of the internet (dynamic routing, etc..):
If www is located far away from www1 and www2, those mon checks may fail,
although www1 or www2 may be accessible for the customer. Other way round:
the mon checks might not fail, but the customer is not able to connect. For
this setup, the machines should be close together (spoken in terms of net
connectivity).
> 4. ALL incoming HTTP requests are initially directed to www which is
> always available. one only need issue this URL to one's clients.
> 5. www simply issues a HTTP header redirect either to www1 or www2
> according to their health/ redirect count.
> 6. after the redirect the HTTP client continues to access the server
> it was redirected to.
And then the customer bookmarks this server (www1 instead of www) and uses
it always. :)
If you need redundancy between two datacenters, you need two multihomed
centers and put 2 heartbeat-connected webservers (or lvs-directors) in each
one. Then put round-robin dns on both datacenters, or work with LVS-TUN.
Regards,
Bjoern
|
