Open Source Insider

MapR ecosystem pack amplifies Kubernetes connections

Adrian Bridgwater
Adrian Bridgwater

Data analytics firm MapR Technologies has sealed the cellophane on the MapR Ecosystem Pack (MEP) at its 6.1 version iteration.

The toolpack is meant to give developers (and data scientists, unless they happen to be the same person) flexibility in terms of how they access data and build AI/ML real-time analytics and, also, flexibility for building stateful containerised applications.

MEP 6.1 also expands on the Kafka ecosystem, adds new language support for the MapR document database and support for Container Storage Interface (CSI).

“MapR was first to solve the stateful container challenge – first with Persistent Application Client Containers (PACC) for Docker containers, then with Flex-volume driver for Kubernetes,” said Suzy Visvanathan, director, product management, MapR.  

Visvanathan says this release is all about helping developers achieve greater independence between Kubernetes releases and underlying storage.

The CSI Driver leverages MapR volumes to provide a scalable, distributed persistent storage for stateful applications — and so this means that storage is no longer tightly coupled or interdependent with Kubernetes releases.

“Implementation of CSI provides a persistent data layer for Kubernetes and other Container Orchestration (CO) tools, such as Mesos and Docker Swarm,” said Visvanathan.

Released on a quarterly basis, MEPs are intended to give users access to the latest open source innovations in this space.

The company says that MEPs also ensure that these updates run in supported configurations with the MapR Data Platform and other interconnected projects.

For the MapR Database… there are new language bindings for Go, C# to give developers a chance to build a broader set of new applications in the language of their choice on MapR document database. Existing languages include Java, Python, Node.JS.

There are also Oozie 5.1 enhancements (Oozie is a workflow scheduler system to manage Apache Hadoop jobs) to move dependency from Tomcat to Jetty for embedded webserver, which is much more lightweight (and most would agree secure) and update the launcher of Oozie which is generic to YARN, instead of in the MapReduce format.