Follow Datanami:
April 1, 2020

Cloud DB Combines Kubernetes with Open Middleware

A database service built on the Kubernetes cluster orchestrator takes a multi-cloud approach that allows a master database to run in one cloud while replicated instances are distributed across other cloud services using emerging open source middleware.

YouTube spinout PlanetScale is the company behind Vitess, the open source database clustering system developed by the Google-owned video service. PlanetScale’s cloud-native database is billed as the first built upon Kubernetes and Vitess with the goal of extending MySQL features across the three top public cloud platforms.

The goal is to help customers “liberate their data with a cross-cloud database, making it effortless to switch from one cloud to another and overcome the challenges and costs associated with cloud vendor lock-in,” said Jiten Vaidya, PlanetScales’s CEO and co-founder.

PlanetScaleDB is built on Vitess, which is currently used to run “very large” OLTP workloads for GitHub, Slack and other users, the startup said.

Founded in 2018, PlanetScale is among a growing number of vendors developing cloud-based databases services. Market tracker Gartner (NYSE: IT) notes that cloud deployments of databases have accelerated over the least two years. For example, database vendors such as DataStax have been moving their implementations of Apache Cassandra distributed database to hybrid and multi-cloud platforms.

PlanetScale, Mountain View, Calif., and other cloud-native database vendors also are offering managed services that allow deployments in just a few clicks. The startup differentiates its MySQL offering as built on the de facto standard Kubernetes cluster orchestrator along with Vitess, the database clustering platform certified by the Cloud Native Computing Foundation.

Along with YouTube, Vitess is also used to handle production workloads for the Slack collaboration platform and the Square cash app, among others.

As more databases are migrated and deployed to the cloud, startups like PlanetScale are betting enterprise customers will want multi-cloud safety nets in place to avoid downtime and data loss. Hence, managed database services are increasingly offering failover and disaster recovery options.

Along with scaling, the Vitess clusterer is designed to automatically handle outages and backups.

YouTube began working on Vitess a decade ago to manage a huge number of MySQL instances and the resulting database outages. “What is has now become is basically…full-fledged sharding middleware for MySQL,” said PlanetScale CTO Sugu Sougoumarane.

Without Vitess partitioning of a MySQL instance, “you’d have to make pretty big changes to your app if you want to shard your database,” Sougoumarane added.

Those capabilities were eventually combined with Kubernetes to make PlanetScaleDB a cloud-native option for distributed databases.

PlanetScaleDB is generally available on Amazon Web Services (NASDAQ: AMZN) and Google Cloud Platform (NASDAQ: GOOGL), with support for Microsoft Azure (NASDAQ: MSFT) added this week. Gartner reported last June that AWS and Azure accounted for the lion’s share of cloud database growth.

Recent items:

Cloud Now Default Platform for Databases, Gartner Says

DataStax Unveils Constellation, Its Cassandra Cloud Platform

ScyllaDB Gives Cohabitation of OLAP, OLTP a Shot

-Editor’s note: This article has been updated to note that some customers have yet to migrate to PlanetScale’s new multi-cloud service.