Sign up for the ITPro Today newsletter
Stay on top of the IT universe with commentary, news analysis, how-to's, and tips delivered to your inbox daily.
[Root] Access is an advice column for questions about IT issues, career moves, and workplace concerns.
Submit questions anonymously using this form.
As an IT admin managing a hybrid cloud storage environment, I recently dealt with prolonged downtime because our primary cloud storage provider suffered an outage. The outage was a huge disruption. Unfortunately, our SLA [service-level agreement] didn’t cover us adequately. I’m now reevaluating our reliance on a single provider and considering options to reduce this risk.
How can I design our storage architecture to be less dependent on a single provider’s availability? Should I look into a multi-cloud storage strategy, or are there other approaches to enhance our cloud resiliency?
—No More Cloud Catastrophes
If you prefer sticking with a single cloud provider, configuring your workload for regional failover is the best way to protect against outages. Most cloud providers offer options for workload replication across regions for resilience, allowing failover if one region becomes unavailable. However, for this to work effectively, you must set up a secondary region to actively run your workload rather than merely serve as a backup.
That said, regional replication alone may not protect against all types of outages. Some incidents can impact multiple regions within the same provider. You may be better off adopting a multi-cloud strategy since an outage won’t likely affect multiple cloud providers simultaneously.
If you choose a multi-cloud deployment, distributing resources across regions is still a good idea. For example, if your workload is on the U.S. East Coast in an Azure data center, consider replicating it to an AWS data center on the West Coast. The geographical diversity allows your workload to continue running if a regional disaster, like a major hurricane, disrupts one location.
Although multi-cloud may be your best option, it’s not the only option. You could also consider a hybrid strategy. In this setup, you would host your primary workload in the public cloud while replicating data to servers in your data center. It allows you to temporarily run the workload in-house if the cloud provider experiences an outage. Once the provider has restored its services, you can fail back to the cloud, resuming operation in the cloud data center.
Whether you adopt a hybrid or multi-cloud strategy, ensure you enable two-way data replication. Here's why: Suppose you host a workload in the cloud but replicate it in an alternate location, either in another cloud or your data center. If the cloud provider has a long-duration outage and you shift the workload to the alternate site for a few hours, it will accumulate data during that time. When the cloud provider restores its service, you must fail back to it. Without two-way replication, your cloud data will be outdated and miss any transactions that occurred during downtime. Two-way replication ensures data accumulated in the alternative location syncs back to the cloud, keeping your cloud-based data updated and complete.
With a multi-cloud or hybrid strategy, automated failover may require third-party software due to reliance on resources outside your primary cloud provider. If your cloud provider goes down and you lack automation, you might need to transition the workload to its alternative location manually.
As previously noted, multi-cloud or hybrid strategies are viable options for adding redundancy. They allow your workloads to remain operational even if your primary cloud provider fails. However, the two approaches come with associated costs.
For a multi-cloud approach, you will incur costs for the resources consumed in the secondary cloud.
With a hybrid setup, you must budget for the infrastructure required to host a copy of the workload in your data center.
Additionally, data egress fees can add up. Cloud providers typically charge fees when data leaves their cloud. While some may waive these fees for backup or recovery, replicating data to an alternative location will likely incur data egress charges.
Sign up for the ITPro Today newsletter
Stay on top of the IT universe with commentary, news analysis, how-to's, and tips delivered to your inbox daily.
When you subscribe to the blog, we will send you an e-mail when there are new updates on the site so you wouldn't miss them.
Comments