Expect the unexpected with a solid cloud DR strategy
A comprehensive collection of articles, videos and more, hand-picked by our editors
Disaster Recovery as a Service (DRaaS) is an up-and-coming service that can nicely round out a cloud provider's...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
product portfolio. DRaaS is a low-cost alternative to expensive and often unwieldy traditional disaster recovery options. But cloud disaster recovery services are more than just cloud storage; they encompass planning, process, integration, testing and constant vigilance. Done wrong, DRaaS can spell disaster for your customer's business and your reputation. Here's how to avoid the pitfalls and successfully enter the DRaaS market.
Virtually none of the vendors who sell their wares to service providers knows anything about offering DR services on a multi-tenant cloud footprint.
Who knew what a derecho was until this unusually violent type of storm walloped Northern Virginia and resulted in a serious Amazon Web Services outage for its Elastic Compute Cloud 2 (EC2)? The event brought down many of Amazon's big-name customers, including NetFlix, Dropbox, Pinterest, Instagram and Heroku, along with a long list of other unfortunates. The message is that any enterprise whose business interests would be seriously harmed by a service disruption is a potential candidate for DRaaS.
Who wants cloud-based disaster recovery services and why?
Traditional disaster recovery services replicate application state between two data centers, so if the primary data center becomes unavailable, the backup site can take over and activate a new copy of the application using the most recently replicated data. Disaster recovery services are well suited for the cloud because under normal conditions, minimal resources are needed to synchronize state from the primary site to the cloud. The full complement of resources required to actually run the application need only be provisioned and paid for when disaster strikes. Additional resources can be rapidly called upon for quick recovery after a failure, enabling business continuity.
Now, cloud-based DRaaS is a viable alternative to traditional disaster recovery if a customer's applications are already running in a virtualized environment. Applications running on mainframes and hybrid computing environments are best served by traditional disaster recovery offerings.
SunGard, which has been offering traditional data disaster recovery services for more than 30 years, added cloud-based disaster recovery last year. According to Ram Shanmugam, senior director of product management for recovery services at SunGard Availability Services, about 20% of SunGard's customers run their applications in fully virtualized environments -- and that segment now uses SunGard's DRaaS offering. Shanmugam predicts that the percentage of customers with fully virtualized environments will grow anywhere from 20% to 60% over the next two years. If his prediction is accurate, DRaaS will quickly come into its own.
How to implement DRaaS and avoid the pain points
Implementing DRaaS solutions entails creating a cloud disaster recovery plan with Recovery Point Objectives (RPOs) and Recovery Time Objectives (RTOs) based on business considerations and well-defined processes for recognizing and declaring a disaster. It also entails tight integration between the network, firewalls and load balancers, and the Web, database and storage tiers. Storage is at the bottom of the stack, and, according to Shanmugam, DRaaS must be equipped to recover the entire stack above storage or the recovery will fail.
A thorny DRaaS implementation issue is that all operating system software for the original site and the backup site must be monolithic in terms of vendor and version number. This makes it challenging to provide DRaaS offerings from a customer's private cloud to your cloud or between service provider clouds. Any time software changes, incompatibilities can be introduced and things can break.
More on disaster recovery
DRaaS definitions you need to know
Virtual servers, desktops and storage can simplify DR
CloudCast Weekly: Is DRaaS an easy sell with SMBs?
Another implementation problem is that vendor solutions to date have been designed for enterprises and lack a feature set upon which to build cloud-based, multi-tenant solutions. As a result, cloud providers have been forced to do their own software development.
Craig McLellan, CTO of Hosting.com explains the problem. "There is a school of hard knocks involved [in implementing cloud Disaster Recovery as a Service]. One of the key things we've learned is that virtually none of the vendors who sell their wares to service providers knows anything about offering disaster recovery services on a multi-tenant cloud footprint. The APIs available from vendors are woefully inadequate. They don't expose the hypervisor to the customer, so we have to write the software to provide access to the customers to test and administer the service. None of the vendors contemplated that. If they had an API, we could create multi-tenant abstractions -- without that, it is very difficult."
McLellan also points out that vendors are designing products for what disaster recovery used to be. In the past, testing happened annually; RPOs were not as aggressive. "Now we need to test more often, and to do that, the testing must be 100% nonintrusive. Currently we have to do it manually, and that is intrusive. What is needed is a reliable orchestration layer in the middle," he said.
Good performance data isn't readily available just yet, McLellan continued. For a service provider, service-level agreements (SLAs) are key -- therefore performance is key. "Without good instrumentation, we can't offer SLAs. God help you if you can't tell how your service is performing and a customer declares a disaster. It can easily destroy a service provider's reputation."
He said his company overcompensates by putting in its own instrumentation or by putting a person on it because service providers don't get a second chance. "There is no forgiveness in this space. There's lots of competition, and the switching cost is relatively low," he said.
Partnering for more effective DRaaS offerings
So if you plan to jump into the DRaaS market with a do-it-yourself solution today, be prepared to blaze your own trail. If tackling that option may be low in your priority queue, however, you may be able to enlist a partner to help jumpstart your DRaaS market entry.
Given that it is not easy to roll out your own DRaaS offering, you may want to consider outsourcing to a third party. SunGard resells its DRaaS to cloud service providers under the SunGard brand, which has caché in the disaster recovery space. But you can white label the service if you choose.
You can also work with a cloud system integrator like CloudOps. According to Ian Rae, CEO at CloudOps, "We can architect a cloud-based disaster recovery solution and manage it on behalf of a customer. The advantage of working with a cloud system integrator like us is that we have a strong understanding of what is involved in restoring an application to service after an outage."
The future of cloud disaster recovery services
Cloud disaster recovery services are likely to catch on quickly, and forward-thinking cloud providers should keep their eyes on the horizon to take advantage of what might come next. One possibility is service provider diversity. Just as geographic diversity is important for disaster recovery planning, service provider diversity is also important. In preparation, look for ways to provide DRaaS to customers of other cloud providers, which would help you expand your customer and revenue base and possibly win over additional service business in the long run.
Continue reading part two: Brush up on disaster recovery definitions.
Rebecca Wetzel is a principal with NetForecast and also is president of the marketing consulting firm, Wetzel Consulting LLC. She provides data communications industry insight and helps vendors and service providers develop successful marketing strategies.
Jessica Scarpati is the site editor of SearchCloudProvider.com. Share your thoughts on this column via email or by commenting below.