How to overcome multi-cloud disaster recovery challenges
As multi-cloud environments proliferate, data backup and recovery tech is essential. Here's how to assess enterprise multi-cloud disaster recovery capabilities -- and limitations.
As enterprises continue to spread their data across cloud platforms, backup and disaster recovery grows increasingly complex. CIOs must find a way to balance the cost savings they enjoy from the cloud tech with adequate protection from growing threats like malicious hackers and unintended deletions.
A growing ecosystem multi-cloud disaster recovery services are emerging to help, but it's important to evaluate these through the lens specific enterprise goals.
"Multi-cloud is the latest cloud-related buzzword," said David G. Hill, principal and founder Mesabi Group, an IT analyst firm. "It implies that an on-premises private cloud interfaces with one or more off premises clouds provided by third-party vendors."
The step in cutting through the hype is concentrating on the type service that you want, such as disaster recovery, or DRaaS, when making the decision, Hill said.
But because there are so many different kinds multi-cloud scenarios to consider, it is vital to test recovery processes to make sure the service works when it counts. It's also important to assess latent cloud disaster recovery limitations: Hill and others said to be prepared to adopt or develop new disaster recovery tools as the technology and services improve.
Different DR use cases
There are a few ways customers interpret backup and disaster recovery for the multi-cloud world, explained Naveen Chhabra, principal analyst at Forrester.
- Backing up the cloud service from the cloud source itself.
- Backing up the traditional on-prem infrastructure.
Tools for backing up on prem data are more mature than for backing up SaaS apps, however, creating a tough gap to fill.
Another big problem is that few vendors deliver a solution with a strong policy-based engine, particularly for cloud apps. This means more work for backup admins as the environment grows. This has fueled the growth new vendors like Rubrik, Cohesity and Reduxio that are challenging incumbents with better tools for backing up from cloud apps with some policy and governance capabilities. These are more advanced than traditional disaster recovery apps for on-premise apps.
Prepare for different scenarios
The cloud makes several processes easier: It's easier to move data and connect it to apps. It also makes it easier for rogue employees or hackers to do bad things -- including deleting data. The trick is to bring the same level protection as traditional on-prem disaster recovery to the cloud. For example, the ticket service Live Nation implemented a multi-cloud disaster recovery solution to help back up 120 applications it recently moved to the cloud.
Jake Burns, vice president cloud services at Live Nation, looked at using native cloud backup services like AWS Glacier that would have worked well for physical protection but would have left the data exposed to hackers or rogue employees with AWS credentials.
"They could delete the production and backups as well," said Burns. "You don't want all your eggs in one basket."
Burns advocates against keeping the primary and backup data in the same account to prevent the same credentials or individuals from having access to both. A multi-cloud disaster recovery strategy creates a stronger firewall between the primary data and backup, but costs more than doing everything on one cloud platform.
"You can get 95% the benefit with far less complexity using just one cloud provider," Burns said, but added that the peace mind that comes with the multi-cloud environment was worth it. Every CIO has to weigh the tradeoffs between a slightly higher level of reliability and the additional costs for running across multiple clouds.
Test recovery scenarios
The real test of any disaster recovery service lies in ensuring that it is easy to roll backups into enterprise production. Although the testing process can be difficult and time consuming, it is better to incur an outage during a maintenance window rather than as a surprise after it has failed.
"Everyone implements backup, but few test it," Burns said.
Jake Burnsvice president of cloud services, Live Nation Entertainment
There are two strategies for doing this: Full recovery from a backup, or maintaining a second running system which can be rolled over to when required. The latter approach makes it easier to do QA and testing on this copy and, if successful, to promote it to a primary application.
"If you have the DR system in place, you can count on it because you are periodically switching between them," said Burns.
This same strategy can also be used for cost optimization to take advantage of price fluctuations between different regions. In theory, this would also make it possible to migrate across cloud platforms, but this would also increase operational complexity.
Focus on trust and simplicity
Finding a vendor or vendors with whom you are comfortable with and have great confidence in is critical, Mesabi's Hill said. This means developing an ongoing and evolving checklist requirements and using it to compare multiple vendors.
"But you must be careful because vendors are good at getting checkmarks," he cautioned.
Going forward, analytics-driven policies could become a major differentiator between disaster recovery providers, Chhabra said. Enterprises could develop these policies based on data allocation, sensitivity and recovery point objectives.
"This is a clear application analytics for the backup space, unfortunately, none the vendors are delivering such insights," Chhabra said.