Business Continuity

Business Continuity is the art of making sure that a business can continue, even in the face of natural or man-made disasters such as fire, flood, hurricane, tornado, or bombing.

There are many approaches to this subject, but the one that interests us the most is the one which treats disaster recovery like HA over long distances. In other words, the backup site has the resources to take over from the main site hot with data replicated up to the second with a technique like DRBD.

As an overview of this approach, one might like to read Christoph Mitasch's thesis on Server-Based Wide-Area replication for Disaster Recovery.

It's a reasonable introduction to the topic.

Business Continuity in Release 1 Configurations

As far as using Linux-HA release 1 in a business continuity situation, the only thing we can offer right now, is to use it with DRBD (or other replication), and then use the meatware STONITH device, and a longish deadtime (like a few minutes).

This means that takeover will never occur until a human concurs with the decision to take over things.

The disadvantage of this approach is that it requires that when you lose communications for longer than deadtime that you will have to restart heartbeat on the slave side, in order to make it stop trying to STONITH the "live" side, and start monitoring things for another takeover again.

It's far from optimal, but it may be workable in some situations.

Business Continuity in Release 2 Configurations

CRM-style configurations in release 2 will eventually have significant advantages in terms of business continuity type configurations. A proposal for these new features can be found at the SplitSite page.

Linux-HA site built by the Linux-HA crowd. HTML4 / CSS2