Building a disaster recovery architecture with cloud and colocation
A comprehensive collection of articles, videos and more, hand-picked by our editors
Any discussion of data center disaster recovery will almost certainly involve a discussion of off-site facilities....
Businesses can select a wide scope of disaster recovery facilities, ranging from traditional, nonhosted colocation to managed colocation services to full cloud-based services. The actual choice depends on the particular needs and capabilities of the business. As disaster recovery options proliferate, pay more attention to disaster recovery requirements, compliance and testing on an ongoing basis.
One of the most challenging and complex aspects of IT planning is preparing the data center and the business to handle inevitable disasters. Traditional disaster recovery methods like remote backups and replication are still in play, but IT organizations now look outside the local data center to better guard themselves against serious facility damage. Colocation has emerged as an attractive option because a business can essentially replicate the primary data center offsite. The disaster recovery colocation facility can step in with little or no disruption when trouble strikes.
But colocation isn't a quick or simple endeavor. Let's address two vital questions when considering disaster recovery colocation.
Q. What should I look for in a disaster recovery colocation facility?
The bottom line is that a colocation choice must be "available." Here are some factors for evaluating a potential colocation provider.
Consider the physical location, which should be stable -- geologically, politically and economically. If you plan to deploy hosted server colocation, where equipment and staff are provided, take into account the IT talent available in the region, such as local technical colleges and other IT industry in the area.
Evaluate the accessibility and resource stability of the colocation facility. For example, look for multiple independent power sources, reliable long-term on-site power generation and multiple independent connectivity options, such as telcos and internet service providers, to service the building. If the building relies on water from a river or lake for cooling, check that the water source is stable year-round.
Don't overlook security. Your business data is always on the line whether it's your physical hardware or not. This means the colocation provider should have ample security measures in place to restrict and monitor access to the building and the hardware areas.
Know who will answer the phone when you call, and understand the process of escalation needed to resolve issues quickly. The finest colocation facility might be inadequate if you're always leaving voicemail messages and your email messages go unanswered. Look at the provider's ownership history and see how often the business has changed hands. Each management change will inevitably bring policy, service and cost changes to your business and other colocation clients, so long-term management stability can be a positive sign for the provider.
Take into account the human and social factors in disaster recovery support. If you do business with a conventional, nonhosted colocation provider, chances are you'll have equipment and staff on-site with the provider from time to time. In a disaster situation, that staff may be the only team that's actually keeping your business in operation.
While many of the issues above would be considered the responsibility of the provider, that vendor is your business partner. Anything that can disrupt the colocation provider will threaten disruptions to your disaster recovery preparedness and possibly your regulatory compliance posture. No colocation provider is going to supply compensation for lost business other than a credit for any downtime hours, so the onus is always on the customer to perform due diligence and investigate the long-term physical and technical resources of the vendor.
Q. How close should a colocation provider be to my main data center?
The answer to that question depends on the disaster scenarios you want to overcome. If the idea is to guard against local facility damage from fire or flooding, for example, you can select from numerous economical colocation options just outside of the danger area. However, if you're trying to guard against a large-scale disaster, your disaster recovery colocation site might need to be selected somewhat farther away from your main facility.
Workload performance requirements and staff travel logistics often dictate colocation distance decisions. Distance imposes signal latency, so the farther away a colocation site is located from the main data center, the longer it takes to move data across the network between the local and colocation sites.
Assuming a failover disaster recovery colocation site is in fact farther away from the main users, expect a slower user experience. Network bottlenecks, switching configuration issues and ISP traffic levels can further exacerbate network issues that might occur between the failover site and end users.
Asynchronous and general business applications tend to be less sensitive to latency issues, but many high-end enterprise applications, such as transactional or real-time workloads, can be extremely sensitive to latency. This can make it difficult to synchronize data sets between distant sites, and disruptions might carry a higher potential for data loss. Understand the data handling needs of your business applications and select a colocation site that is distant enough to guard against the disasters you need to avoid -- yet close enough to minimize latency.
Weigh the effect of colocation distance on your IT staff. Unmanaged colocation may require you to dispatch staff to run the colocation site during a disaster, as well as routine equipment installations, replacement and maintenance.
Verizon data center outage grounds JetBlue systems
Colocation with a side of cloud
Ask a colo: What's an Uptime Tier worth?
Data center provider changes colocation pricing scheme