Ensuring the sustainable availability of mission-critical IT systems and data is increasingly expensive, time-consuming and difficult.

New approaches, including cloud computing, support a more flexible approach for more-frequent disaster recovery (DR) testing, and can make IT disaster recovery management less burdensome, but most options are in the nascent stages. They have yet to prove themselves as viable alternatives for production systems and data failover and should be considered only for low-risk deployments. 

Traditional DR approaches impractical

The explosive growth in the number of applications and in the amount of data to be protected -- as well as in recovery expectations -- is making traditional DR methodologies increasingly impractical. 

Today, public cloud providers offer two DR-specific classes of infrastructure-as-a-service (IaaS): DR "in the cloud" and cloud-based storage services. The providers include both startups and more-mainstream recovery and hosting service providers introducing first-generation offerings. 

The maturity of their offerings varies in how well they address three key criteria: infrastructure reliability, ease of use, scalability and manageability. Several IaaS providers already offer cloud-based solutions, and some providers of traditional, whole-site recovery services have recently introduced formal cloud-based solutions as well. 

Dramatic cost reductions

Cloud providers can present very inexpensive alternatives to traditional DR providers when dedicated services are not required and when application scope is relatively limited. 

Cloud computing can cost as little as US$0.5 per processor core per hour, and cloud storage as little as US$0.1 per gigabyte per month. These rates compare favorably with monthly subscription fees as high as US$150 per Windows server per month, and US$1 to US$4 per gigabyte per month from more-traditional DR service providers. 

The potential customer benefits of cloud-based DR services vary, depending on a number of factors. These include the diversity of the computing platforms that require recovery support; the customer's ability to orchestrate, and ideally automate, recurring recovery testing; and the customer's ability to transparently and efficiently use cloud-based storage for ongoing data backup, replication and archival. 

In the near term, cloud-computing and storage services for Windows- and Linux-based applications will be far more pervasive than for other computing platforms, such as IBM zSeries mainframes and iSeries midrange systems, and vendor-specific Unix platforms, such as Oracle Solaris, HP-UX and IBM’s AIX.