Cloud computing is taking the planet by storm. Many businesses are taking advantage of this low cost and easy-to-access option. OPEX billing model and easy scalability are one the most important benefits of working in the cloud, however there are more valuable gains from cloud. The highly advantageous option is top-notch disaster recovery. Cloud computing is making the disaster recovery process easier, faster, and more cost-effective.
In this article I will discuss about two most popular cloud platforms : Amazon Web Services (AWS) and Microsoft Azure, and their capabilities to provide disaster recovery.
- Microsoft Azure has a dedicated service for disaster recovery called “Site Recovery”.
- Site recovery can be used to setup DR for on-premise physical and virtual environment. Hyper-V and VMware environment is supported.
- Its agent based DR setup where you deploy agent on the host and on VM’s ( for VMware), replicate it to Microsoft Azure. Site Recovery also has orchestration components which lets you decide the order in which VM’s should start in Azure. For example : you can configure to start AD first, then web front end, followed by application server and at the end the database server.
- Site Recovery can help organization to failover to cloud and failback to on-premise from cloud. NOTE : when you failover a physical server to Azure, via Site Recovery, you can only failback to an onpremise VMware infrastructure.
- Replication frequency – least is 30 seconds followed by 5 minutes and the larger value is 15 minutes.
- You can perform Test fail over into an isolated cloud virtual network without impacting your on-premise network/server infrastructure.
- One of the important features of Site Recovery is that VM’s are not up & running when you replicate them from on-premise. Only when you failover to Azure, the VM’s are started and user requests are catered for. This saves a lot of money to customers.
- From a security standpoint, it supports encryption at rest and encryption in transit.
- Application Awareness – It knows what you’re running (i.e. SharePoint, SQL Exchange, Active Directory, etc.) .
- Region to Region Replication – If you want to take your replication from the East Coast to the West Coast, this is built into the service, so it’s easily done.
Amazon Web Services
- AWS combines several components such as Route53, EC2, S3, CloudFormation, VPC, Storage gateway etc.. using which a DR solution can be created for any infrastructure (physical/virtual). AWS does not offer a dedicated offering for disaster recovery.
- We can achieve DR in AWS using one of the following ways : Backup/Restore, PilotLight deployment, Warm standby, Hot Standby.
- Backup/Restore : We move data to AWS using AWS Import/Export service or AWS Staorage Gateway. Store this data on S3. Create VM using pre-configured image (AWS Machine Image), attach S3 to this VM and you have the VM running in AWS.
- PilotLight : Setup the infrastructure in AWS similar to your on-premise, but do not start/run web front end and application servers. Enable database replication from on-onpremise to cloud via VPN. Below is a reference architecture.
During a disaster, Route53 will redirect traffic to AWS. CloudFormation will be used to start the VM’s and cater to web requests.
3. Warm Standby Deployment : Similar to Pilotlight, in Warm standby we will setup the infrastructure in AWS similar to your on-premise. Difference here is that the VM will be up and running on a lower configuration spec. VM. In case of a disaster, Route53 will divert traffic to AWS, using CloudFormation we can scale up the VM’s to handle production load.
4. Hot standby (multi-site deployment) : Similar to Warm Standby, in Multi-site deployment we will setup the infrastructure in AWS similar to your on-premise. Both the environment will be identical with respect to configuration. You can use Route53 to distribute user request between on-premise and AWS. In case of disaster, user requests gets redirected to AWS. Below is a reference diagram.
- Applications deployed on AWS have multi-site capability by means of multiple Availability Zones. Availability Zones are distinct locations that are engineered to be insulated from each other. They provide inexpensive, low-latency network connectivity within the same region.
- Failback option is easier and simpler in Pilotlight, warm stand by and multi-site deployment when compared with backup/restore in AWS.
Based on this information, my personal opinion is Microsoft Azure is better suited for cloud based Disaster Recovery. It covers most of the DR concepts and has good integration with on-premise virtual infrastructure.
You can also look into Veeam, Veritas VRP , DoubleTake, Commvault solutions with Azure/AWS to build a DR solution.
Hope this article gave you enough information to select suitable cloud platform.