Disaster recovery

The Common Web Platform (CWP) quickly responds and recovers to another location in New Zealand if a major disaster — like an earthquake — takes the primary data centre offline. You can pay for 3 different recovery options depending on how quickly you need your website up and running.  

1. Backup-only


Cost

Included with the production environment. You can choose to backup user acceptance testing (UAT) and development/testing environments for a small extra monthly fee.


How it works

It’s based on a daily backup of your website, which is stored offsite. You could lose content and data changes since the last backup.


Recovery time

Sites will be restored, but there’s no service level or guarantee on recovery time.

2. Passive disaster recovery


Cost

A small extra monthly fee.


How it works

A copy of your website is continuously mirrored from the primary to the secondary data centre. This reduces potential data loss and increases recovery speed. In the case of a disaster, we will ‘boot up’ virtual machines and make networking changes at the secondary data centre to restore the site.

You can only get passive disaster recovery for your production environment — not for UAT or additional test and UAT environments.


Recovery time

20 hours (Contracted Service Level).


3. Active disaster recovery

Diagram showing how active disaster recovery works


Cost

A larger extra monthly fee and a further fee if your UAT environment is protected (which is recommended).


How it works

The environment (the CMS application, files and database) is actively replicated between the primary data centre and the secondary data centre. Website traffic is served from both data centres by using a geographic load balancer provided by Incapsula.

If either one of the 2 data centres fail, the system responds by serving traffic from the working data centre only. This reduces the potential for a website outage.

A third node acts as a ‘witness’, communicating with the databases in the primary and secondary site about whether they should operate.

The geographic load balancer itself is also actively replicated in multiple locations, so if the geographic load balancer fails, it can automatically work from another location.


Using active disaster recovery for your UAT environment

We recommend using active disaster recovery for both your UAT and production environment. If you do, it means that in a disaster your UAT and production environments will have the same server configuration. You’ll be able to:  

  • accurately test production performance and locate issues  

  • write and test code that works in the production environment, preventing the chance of an outage

  • quickly develop and deploy code.

Recovery time

4 hours (Contracted Service Level). However, load balancing is automatic and can work in a much shorter timeframe — usually minutes.

Last modified: