We asked some of the hundreds of cloud professionals in attendance at our recent RightScale Compute conference for their perspective on key cloud computing issues. On the topic of high availability (HA) and disaster recovery (DR), we saw several themes emerge:
- HA and DR in the cloud have matured significantly.
- Experience in the cloud environment helps organizations architect enterprise-grade HA and DR.
- Depending on the use case, there are different best practices for how companies architect HA and DR.
Dave Nielsen, principal consultant with Platform D, said, "It used to be that enterprises looked at high availability and disaster recovery as two different solutions. More and more are looking at the cloud as a good solution for both."
Steve Martin, Microsoft's general manager for Windows Azure, was among those who said, "The cloud, itself, could be a disaster recovery environment." He cited running backup copies in the cloud "that are ready to go if something happens on the premises, or extended high availability for peak load capacity."
How are high availability and disaster recovery improving in the cloud?
To that point, Peder Ulander, vice president of product marketing for cloud platform at Citrix, said, "We've seen companies that are building private clouds and tying them to public clouds. These things are portable, encapsulated, and designed in such a way that you can replicate out to a public cloud or back in from a public cloud, so that when disaster hits, your service and your availability is still there."
Justin Pirie, cloud strategist at Mimecast, agreed: "I'm seeing cloud services enabling much higher levels of availability than on-premise services." As an example, he noted that his company replicates three copies of all data and replicates it in real time across two sites.
That's necessary, because, as Vanessa Alvarez, head of marketing at Scale Computing, noted, public cloud service providers have had some high-profile outages over the last couple of years, but she said they have learned from those episodes and have "evolved their architectures to become much more resilient and reliable for the enterprise IT customer."
Duke Skarda, CTO of SoftLayer (recently acquired by IBM), said that initially his company's customers "expected cloud providers to do HA for them. As they leverage the cloud more and more, [they] understand that they're still responsible for understanding their application and making it resilient in the new environment the cloud provides, and leveraging the cloud tools in a new way that they couldn't do when they had their own ... data center sites."
Brian Goldfarb, head of cloud platform marketing for Google Cloud Platform, concurred: "A lot of high availability and disaster recovery is about how you design your application." He cited the use of multi-homing and multi-zone architecture.
Of course the HA/DR benefits that enterprises get from the cloud require some planning. Chris Henry, head of technology at Behance, says one difficulty of recovery "is being able to replicate all of your tools from one side to the other. Some of the tools that are coming out that allow you to wrap LVM [Logical Volume Manager] and move a large chunk of data from one place to another [will let people] take down an existing infrastructure and just pop it up somewhere else with minimal downtime to users."
In addition to the industry leaders cited here, we also spoke with a number of other cloud experts who shared their insights in this six-minute video.
To get advice on how to architect for HA and DR, request a free consultation from a RightScale cloud expert.