Disaster Recovery in the Cloud

Most small and medium sized business do not have a formal Disaster Recovery (DR) plan and implementation because of its cumbersome and costly nature. Various factors make DR complex, including: (1) Allocation and administration of remote compute and storage resources; (2) Data transport mechanism - e.g. tape shipment or data replication; and (3) Application environment synchronization. To makes matter worse, regular testing of a DR implementation tends to be complicated, and in many cases not practical.

Cloud Computing provides an excellent means to radically simplify the DR process. This is achieved by backing up your critical applications to a Storage Cloud (e.g. Amazon S3), and making preparation to quickly recover in the nearby Compute Cloud (e.g. Amazon EC2).

We have two solutions for backup and DR in the cloud: Amanda Enterprise (with the Amazon S3 Option) and Zmanda Cloud Backup (ZCB). Amanda Enterprise is meant for environments with heterogeneous systems, whereas ZCB is targeted at small businesses with a handful of Windows servers and desktops.

Amanda Enterpise DR in the Cloud

Setup of Amanda Enterprise for Cloud Based DR

 

Zmanda Cloud Backup DR in the Cloud

Setup of Zmanda Cloud Backup for Cloud Based DR

 

The process of setting up DR in the cloud is as follows:

  1. Set up backup process to Amazon S3.
  2. Complete first backup of applications on primary site to S3.
  3. Configure standby VMs on EC2 to match the OS (and patch level) of the corresponding systems on your primary site. For all data storage, use Elastic Block Storage, so you have persistent data across reboots.
  4. Install Zmanda backup software on these standby VMs.
  5. Install the same S3 certificate that is used in step #1 on the standby VMs.
  6. In case of Amanda Enterprise setup the AE-DR option to replicate backup catalog and configuration to the standby VM running the AE server.
  7. Perform full recovery from S3 to standby VMs.
  8. Take a snapshot of the standby VMs.
  9. Shutdown standby VMs.
  10. Optionally start standby VMs periodically to perform steps #6-#8. This will help in reducing the time to recover after a disaster and also tests your DR process.

If you are considering the Cloud for your DR needs, come join us tomorrow (June 22nd) for a webinar: Noted Storage Analyst, Lauren Whitehouse from Enterprise Strategy Group, will be joining me: Leveraging the Cloud for Radically Simple and Cost-Effective Disaster Recovery

Comments are closed.