No matter the size of the disaster, user expectations regarding the speed and seamlessness of the recovery effort have increased dramatically. While IT budgets have not increased significantly to meet these beliefs, there are technologies that can help. Here are five steps IT professionals should consider to cut disaster recovery costs when designing or updating their DR strategy.
Step 1: Understand the recovery need
The most important way to reduce disaster recovery costs is to understand what needs to be recovered in the event of a total data center outage. This can be achieved by establishing a recovery point objective and recovery time objective for each application. However, it is important to consider unstructured data in the analysis.
Though most data centers are awash in applications, approximately 5% of them are mission-critical. While these applications tend to require a small amount of the overall storage capacity of the data center, they are the most demanding from a performance perspective.
Step 2: Correctly size the DR storage system
When IT professionals size the storage system for the disaster recovery (DR) site, they often assume it must have the same or similar capacity and performance as the primary system. In most situations, this is not the case. In a real disaster, only the mission-critical applications need to be recovered quickly, and they only need access to the latest copy of data.
If the IT team has done the front-end assessment work, it can leverage this knowledge to keep disaster recovery costs down. Since users only need immediate access to the documents they were working with most recently -- typically a two-week timeframe -- the capacity of the storage system at the DR site can be dramatically smaller. The result is that a 100 TB storage system in the data center can now be less than 10 TB at the DR site.
The DR system needs to provide similar performance to these applications. If the primary data center used flash storage for these applications, the DR site should at least have a flash tier acting as a cache. It is important to emphasize that, even though these mission-critical servers are the most demanding in terms of performance, they still represent a small mixture of the data center workload, so the DR system should not need the same performance as the primary.
Step 3: Count on virtualization
Even if the primary data center is not 100% virtualized, the DR site should be virtualized. Virtualization keeps server costs at a minimum and can further reduce capacity requirements by allowing the DR storage system to support deduplication and compression. Virtualization has another important benefit: It makes testing the DR plan far easier. IT planners can create virtual isolated test networks and enable DR copies of applications without impacting production.
Step 4: Consider disaster recovery as a service
Disaster recovery as a service (DRaaS) takes the concept of a virtualized DR site to the extreme. Cloud backup providers are racing to add DRaaS to their feature sets. Instead of just backing up data to the provider's cloud, DRaaS allows an organization to also instantiate virtual machines in the provider's cloud, borrowing its compute to enable rapid recovery. DRaaS saves organizations the cost of building a separate DR site, as well as the expense of equipping that DR site with servers, storage and personnel.
For small and medium-sized businesses that lack a secondary location but have IT personnel, DRaaS is compelling in terms of disaster recovery costs and operations. For data centers with a secondary site, the cost of storing years' worth of backup in the cloud may be impractical. That said, enterprises can steal some of the concepts from DRaaS providers and implement their own form of the technology.
Step 5: Roll your own DRaaS
DRaaS counts on a secondary data center with enough resources to not only store backup data, but to deliver enough compute to run applications in the event of a data center failure. Enterprises can build their DRaaS offering, but have better control over cost containment.
Instead of directing all backups to the cloud and paying to store it, enterprises can replicate the data from their most mission-critical data sets to a cloud provider like Amazon, Google or Microsoft. Several software applications can provide that functionality. Some of the applications will also convert enterprise virtual images into virtual cloud images so the application can run unchanged in the public cloud. This type of offering saves the expense and upkeep of a DR site while maintaining organizational control.
DR expectations are higher than ever. Users want the IT staff to make any recovery happen quickly, seamlessly and for less money. Understanding your applications and data is the critical first step in decreasing disaster recovery costs, minimizing the outlay of capital in the subsequent steps noted above.