Announcing True Multi-Cloud Clusters for PlanetScaleDB
If you are like the majority of engineering leaders, you are in the middle of determining strategies for digital transformation. Kubernetes has emerged as an operating system for compute resources, either on your own data center or in the public cloud. This provides a clear path for managing your stateless applications. Kubernetes makes it easy to migrate stateless applications across your own data center and public clouds.
But data has gravity. And this has implications for disaster recovery and vendor lock in because migrating data across public clouds without service interruptions is a non-trivial feat.
What you need is the ability to deploy, manage and monitor your databases where the master and replicas are distributed across clouds. We have built a database-as-a-service, PlanetScaleDB, on top of popular and battle hardened open source projects: Vitess, MySQL, and Kubernetes, that does just that.
Starting today, you can try out PlanetScaleDB clusters that span regions in a single cloud provider and also across cloud providers. Also, we have added support for Microsoft Azure. We now support 4 regions each for Amazon Web Services, Google Cloud Platform, and Microsoft Azure. We are making these features available as a beta version; try them out in your dev and staging environments and let us know what you think. We plan to make these features GA within the next 90 days.
Why use true multi-cloud clusters?
Two reasons: Disaster recovery and freedom from vendor lock in.
An outage in a cloud provider across the whole region can have catastrophic consequences for a business. The ability to quickly failover to a nearby region, whether in the same cloud provider or different cloud provider, and continue to serve traffic with minimal downtime is critical. Also, once the disaster is over, the ability to migrate back to your most optimal operational configuration with the same ease is important. PlanetScaleDB true multi-cloud makes this possible.
Similarly, when negotiating with a cloud provider, it is always useful to have the option to painlessly migrate out of the cloud provider. With PlanetScaleDB true multi-cloud you can now start replicas in a secondary cloud provider giving you the option of seamless data migration between cloud providers, if needed.
A true multi-cloud database also provides extended geographical reach; if a customer’s primary cloud provider doesn’t have coverage in a given geography, a true multi-cloud database can be deployed with additional replicas in another cloud provider to serve traffic in a specific geographic area.
Even within a single cloud provider, the ability to deploy in multiple regions and seamlessly failover across regions provides significant disaster resilience for a database. By running at least one replica outside of a primary region, a regional outage in the primary region can be handled by failing over to a replica running in the secondary region. Once the outage has been resolved in the primary region, a failover to a replica in the primary region from the secondary region will return the database to the original setup avoiding downtime for end users. This ability to failover back and forth between regions makes PlanetScaleDB multi-region databases especially useful.
Go to https://console.planetscale.com/ to spin up your multi-cloud cluster or read this blog post to learn how to deploy a multi-cloud and multi-region instance on PlanetScaleDB. Read the press release of what we announced here.
The same technology we built for distributing a master and replicas in Kubernetes clusters across regions and cloud providers will allow you to add your own Kubernetes cluster as a custom region. These Kubernetes clusters can be on premise or on a cloud provider of your choice. We call this Bring-Your-Own-Kubernetes or BYOK8s. We will have some exciting announcements coming in the summer so stay tuned for more!