Thursday 31 July 2008

What is the difference between replication, high availability and disaster recovery?

There is a lot of confusion about these terms, and it seems that everyone has their own definition so we thought that we would clarify these terms a little.Before we put label onto anything lets talk a little about what we are trying to do. Here are a number of scenarios:

We want to protect our data so we create another real-time copy somewhere else, this is typically on another server, which is located either locally or remotely
We want to minimise servers downtime, so we want to minimise the time interval between a server being down and a backup server coming back up in its place. The best way to do this is to put both the active and backup servers on the same LAN.
We want to protect our applications, so we want to minimise the time taken for users to re-gain access to their apps. Again the best way to do this is to put both servers on the same LAN.
We want to have the server recovery take place on a remote site to protect against site disasters
We want the application recovery take place on a remote site to protect against site disasters

Scenario no. 1 we call data replication, it requires replication software on both servers.

Scenario no. 2 we call server availability, it requires replication, monitoring and failover software on both servers.

Scenario no. 3 we call application availability and requires everything for Scenario 2 plus application-sensitive monitoring and recovery

Scenario no.4 we call disaster recovery, it will require data replication software, but it may also require a manual procedure for applications to restart on the backup server, and users to be able to reconnect to the disaster recovery site.

Scenario no. 5 is also called disaster recovery (hence the confusion), and it will require replication, monitoring and failover software. But the beauty of this solution is that it can cut down the work required when a disaster occurs such as repointing client connections, restarting the server on the backup site etc.

However, someone who knows what they are doing to oversee a disaster recovery is always recommended.
From our experience, we have found that labels are not as important as knowing what you want to achieve. Having a clear idea of what you want to happen in the case of downtime and having a capable individual to take responsibility for the project is an essential starting point for any successful recovery project.