Skip to main content

How Important is the Cloud for Disaster Recovery and Disaster Recovery for Cloud?

Like cloud computing in general, disaster recovery (DR) can mean different things to different people; however the fundamental principal with cloud DR is that it removes the need for dedicated infrastructure to facilitate a failover.

Tom Brand, cloud solutions service director, GlassHouse Technologies (UK), outlines below the options and considerations for cloud DR.

There’s no doubt that server and storage virtualisation have certainly helped reduce the cost of providing and maintaining a DR solution, however it’s still important to have a dedicated infrastructure of the right size for the business (including servers, storage, racks and data centre space).

The cloud changes everything because, thanks to the technical and commercial flexibility cloud solutions offer, organisations can dynamically scale their DR infrastructure in real time, paying only for the services they actually consume. The combination of virtual servers, low cost storage and the overall agility offered by cloud providers creates a very compelling proposition which is helping IT organisations to further drive down the traditionally high cost of implementing and maintaining a DR solution. It is this combination that makes cloud DR an attractive solution in the Infrastructure as a Service (IaaS) sector.

Currently, there are two key types of DR offered in the cloud. Organisations that are actually running their production systems in the cloud typically have availability service level agreements (SLAs) built into their cloud service offerings, making it the responsibility of the service provider to restore the systems and data.

However, it’s important to note that in the current market these SLAs are often relatively basic with the recovery process having little or no intelligence at the application layer and, because of this, one could question whether it really is DR.

For some, cloud DR is more about using external cloud solutions to provide warm standby systems for your internal applications, where only the application data is continuously replicated into the cloud.

Take the example of a classic three tier web application that requires seven servers in the primary site: when the DR infrastructure and systems are provided out of the cloud, the level of resources allocated depends on whether the application is in Replication Mode or Failover Mode.

During normal operation, the system stays in Replication Mode, and requires only a single low cost virtual machine (VM) whose role is to facilitate the synchronisation of application and configuration data. However, when a disaster occurs, the system enters Failover Mode and at this point the resources required to support the full application are brought online, i.e. seven servers, all of which can be automated.

Moving forward, if more organisations adopt virtualisation and build private clouds underpinned by the same technologies as the external providers, who in turn release their APIs, cloud DR will become much more viable and increasingly integrated. The cloud will essentially become an extension of the live environment so we could potentially see DR become a concept of the past.

Customers need to ask the same questions one would ask a cloud provider when adopting any cloud solution, with a number of questions focussing on their specific DR Service requirements. These questions should include topics such as financial stability, security controls, regulatory compliance, data handling and SLA management.

Once the suitability of the provider has been established, the business then needs to drill down into the technical requirements, such as the DR solution’s ability to provide Database replication and file system synchronisation. Can it provide Dynamic DNS failover and global load balancing across multiple sites?

Do its systems offer 24-hour alert notification to IT teams in the event of a disaster? Primary site monitoring from two separate locations should be offered as a minimum: how does the provider monitor for failures? And finally, for providers who offer managed DR solutions, how often do they undertake verification tests and do they offer verification reporting? Quite simply, you can never ask too many questions.