Internally-hosted CycleCloud instances on a virtualization technology (e.g. VMWare or Xen) should be managed with the virtualization platform’s preferred disaster recovery mechanism. DR strategies for cloud-hosted instances are described below.
Amazon Web Services
Amazon Web Services (AWS) instances can be set up using the synchronization method described earlier in this guide, but that requires the ongoing expense of running a second instance. Instead, you can take regular snapshots of the elastic block store (EBS) volume containing the /opt/cycle_server directory. To use this method, you will first want to create a DR instance with a fixed-address elastic network interface (ENI). Contact firstname.lastname@example.org to get a license key for this instance.
When failing over to your DR instance:
- Create a new EBS volume from the most recent snapshot
- Attach the EBS volume to your instance and mount it to /opt/cycle_server
- Write the license key for your DR instance to /opt/cycle_server/license.dat
- Start CycleCloud with /opt/cycle_server/cycle_server start
The ideal DR setup will have instances in different regions, or at least in different availability zones for maximum resilience to outages. However, environment-specific requirements (e.g. VPN connections) may preclude this.