In this timely post, CloudEndure CEO Ofer Gadish talks about why cloud outages are not the problem and how a multi-cloud disaster recovery strategy is the best way to ensure IT resilience.
In light of recent cloud outages, we sat down with CloudEndure CEO Ofer Gadish to talk about whether the cloud is indeed “tough” enough for enterprise-grade applications and why cross-region and cross-cloud DR solutions are crucial components of a strong business continuity strategy.
Q: People get very nervous when they hear about cloud outages. As the CEO of a company that offers solutions for cloud migration and cloud disaster recovery, how do you respond to concerns about cloud outages?
A: Expecting any infrastructure to be 100% bulletproof is unrealistic. Both cloud and on-premise infrastructures have outages. Running off a cloud infrastructure is not less stable than running on-premise. That said, you should never trust a single infrastructure, cloud or not, for business continuity. At the heart of disaster recovery or business continuity is redundancy – always having an updated, available copy of your applications ready to launch in a separate, independent, and remote infrastructure at any time.
Q: In order to achieve this redundancy, do you recommend utilizing two different regions of the same cloud provider or two different cloud providers?
A: Using two different cloud providers is obviously best because the more independent the two infrastructures, the better. The only scenario in which realistically two separate infrastructures could be down simultaneously is if they are somehow in sync.
Cloud providers do a great job of keeping their regions independent, but the dependency between a Google region and an Azure region, for example, is practically zero, while two regions of the same cloud may have some dependencies. So you don’t want to be in a situation where you replicate from one infrastructure to another, and when one of them fails, it causes the failure of the other one.
Q: Isn’t it a lot of work to set up your disaster recovery using two different cloud providers?
I’ve seen companies that prefer cross-region replication because it seems easier to work with one cloud provider. I can’t say that cross-region replication isn’t a smart move, but cross-cloud replication is better in terms of resilience. And if you have the right tools, cross-cloud replication can be just as easy as cross-region.
Q: What kinds of tools make cross-cloud DR just as easy as cross-region DR?
A: Machine conversion. When you mobilize machines from one region to another within the same cloud, you don’t have to convert machines. But when you move from one cloud to another, different clouds use different hypervisors, different networks, different software. Machines are then incompatible, and you have to convert them. With CloudEndure Disaster Recovery, machine conversion is automatic, it’s on the fly, you don’t have to deal with it. When you failover from one cloud to another, you don’t even have to ask for it, your machines just get converted.
Q: So I have machines on AWS and I want to use Google for my DR, CloudEndure just automatically converts the machines?
Q: Are there other tools that make cross-cloud DR easy?
A: Cloud orchestration. We automatically discover and orchestrate the networks, the firewalls, the configurations, all the things that make up the applications that are neither machines nor data. That makes cross-cloud DR just as easy as cross-region.
Q: Does geography play a role in the effectiveness of cross-region or cross-cloud DR?
A: Yes. It’s definitely important to pick the right geographic location for your DR site. You don’t want it to be too close because you can create correlation. For example, there can be a wide geographic area that has a storm or power outage that affects two data centers. So your two infrastructures shouldn’t be too close.
At the same time, especially when you are replicating across countries, you have to think about legal constraints and regulations. That’s why I recommend using two different clouds in two different geographic locations within the same country.
This reminds me of another important benefit of cross-cloud replication. In some countries, the same cloud only has one region. For example, in Australia, AWS only has a single location. Therefore, if you want to replicate to another public cloud infrastructure in a different region of Australia, you need to replicate to Azure. Likewise, if you are using Google Cloud Platform in Europe and you want to replicate to another region in Europe, Google doesn’t have one. In this case, you can replicate to either AWS or Azure’s European regions, thereby maintaining the same EU regulations but enabling enough distance to ensure resilience.
Interested in learning more about using CloudEndure for cross-region or cross-cloud disaster recovery? Contact us and we’ll be happy to answer all your questions.