This article is more than 1 year old
DataStax updates K8ssandra to help Cassandra operate worldwide
Shares multi-cluster lessons from DBaaS in Kubernetes project
DataStax has released a new open-source Kubernetes operator for Cassandra, the wide-column store distributed database about to work across multiple clusters for the first time.
K8ssandra is DataStax's open-source distribution of Apache Cassandra which is designed specifically for Kubernetes and supports stateful workloads. Released today, the update also supports multiple clusters.
The idea, DataStax says, is to allow users to deploy Cassandra-based applications that require high availability on Kubernetes, across multiple regions.
Patrick McFadin, Apache Cassandra veep at DataStax, told The Register: "This is a big thing that's happening inside the Kubernetes community. We got all excited about Kubernetes, but we're building these islands that have no connections. It's a well-known problem.
"When a Kubernetes cluster gets started up, it takes some explicit work for it to work with another Kubernetes cluster: networking, security, all of those things."
A Kubernetes Special Interest Group has been formed to "programmatically build in this multiple cluster idea and have them all federated, and it's picking up steam," he said.
According to this BMC blog, it is an approach that promises to lower application latency and increase performance.
Meanwhile, DataStax has tackled the problem specifically for Cassandra with its own operator.
- Trouble getting REST APIs into GraphQL? Try our low-code approach, says Hasura
- Enterprise databases deployed in Kubernetes? Proceed with caution, warns seasoned analyst
- DataStax 'pauses' AIOps database project to figure out exactly what AIOps is
- It's not 'Door to Heaven', it's 'Stargate': DataStax reaches out to front-end devs with support for GraphQL
"This has been on our roadmap for a long time because, essentially, Cassandra's success is down to it being multi-data centre," McFadin said. "When we put Cassandra in Kubernetes, it made that impossible or hard to do for operators. We've had some manual tooling around that, but when you're using Kubernetes you want it to be simple. What this does is allows you to set up one Cassandra cluster [to run across] multiple data centres. One cluster around the entire world."
DataStax had a bit of a falling out with the open-source community in 2014 when it pulled back on its involvement in Apache Cassandra.
Although it had launched an operator for running Cassandra in Kubernetes, it found when building Astra, its DBaaS, that there was more to running Cassandra successfully in Kubernetes than just an operator – hence K8ssandra.
K8ssandra is seen as something of a peace offering to the open-source community that fulfils the promise to keep making features DataStax developed for Astra, available in an open-source iteration.
"We have been building an Astra and learning. We've announced a multi-cloud version of Astra and things were learned, as they say. Knowledge has been had, and we're downstream that into the open-source projects," McFaddin said. ®