DataStax Extends Operator Reach for Databases

DataStax this week updated its Operator for K8ssandra, an instance of the open source Apache Cassandra database that runs natively on Kubernetes.

Patrick McFadin, vice president of developer relations at DataStax, says the latest edition of the Operator for K8ssandra will make it simpler for IT teams to provision instances of K8ssandra that are deployed on multiple Kubernetes clusters. It can create a single Cassandra cluster spanning multiple Kubernetes clusters across multiple regions to provide a single control plane for simplified configuration, management and operations.

Operators are, essentially, a design pattern originally defined by CoreOS that makes it simpler to capture the institutional knowledge of an IT team. They implement and automate tasks such as configuration, installation, updates, backups, failovers and restore using software that runs inside the Kubernetes cluster. Operators expose a single object to access pods and services at a higher level of abstraction.

Apache Casandra is a NoSQL database based on a wide-columnar store optimized to process large amounts of data across a distributed cluster. As the building and deployments of cloud-native applications continue to evolve, the number of stateful applications requiring access to data residing on a cluster is starting to substantially increase. At the same time, organizations are also deploying more clusters from the edge to the cloud.

A recent survey from the Data on Kubernetes Community (DoKC) finds 70% of respondents report they are now running stateful applications on Kubernetes clusters; another 20% say they expect to see these types of workloads deployed on the platform. Half of the respondents (50%) note they are now deploying databases directly on Kubernetes clusters to more efficiently access data stored on the cluster.

The survey also finds the primary reasons IT teams are deploying stateful applications in Kubernetes clusters is to ensure consistency (45%), a desire to standardize on Kubernetes (40%), simplify management (39%) and enable developers to self-manage the IT environment (39%). The biggest challenge IT teams encounter when deploying those workloads is a lack of integration with existing tools (35%), lack of interoperability (32%), vendor platforms that only address niche requirements (30%) and a lack of skills (29%).

In general, more IT teams simply want to converge compute and storage management rather than having to hire a separate administrator to manage an external storage system.

McFadin says DataStax is trying to provide a de facto standard operator for managing K8ssandra. Today, there are a handful of operators for Cassandra created by various teams. In some cases, internal IT teams have even created their own operators to manage multiple components of a software stack deployed on a Kubernetes cluster.

Regardless of approach, organizations are clearly looking for a way to manage Kubernetes environments at higher levels of abstraction. Operators are a significant step forward in achieving that goal. However, it’s easy to see how IT teams could become inundated with operators for every type of software deployed on a fleet of Kubernetes clusters. In that sense, operators could quickly become too much of a good thing.