MapR Technologies this week announced it has added support for the Container Storage Interface (CSI) to its platform for running big data analytics applications.
Suzy Visvanathan, director of product management for MapR, says CSI support is critical because it will enable instances of the company’s platform deployed on Kubernetes to access external storage. Developed under the auspices of the Cloud Native Computing Foundation (CNCF),which also oversees the development of Kubernetes, CSI makes it possible for applications running on top of Kubernetes to access data that resides on existing legacy storage platforms, in addition to any data stored locally in the cluster.
That capability is critical to drive adoption of Kubernetes in enterprise IT environments in which most of the data necessary for an application resides in that IT environment. Support for CSI is also showing up in various platform-as-a-service (PaaS) environments that face similar challenges, Visvanathan adds.
MapR is providing support for CSI via version 6.1 of a MapR Ecosystem Pack (MEP), which also adds support for bindings written in Go and C#, a Jetty web server and a plug-in that enables open source Apache Drill analytics software to access Amazon S3 cloud storage. Finally, MEP 6.1 also adds support for a Kafka Schema Registry to enable MapR to integrate more tightly with the streaming messaging platform.
In general, Visvanathan says most enterprise organizations are carefully vetting which types of applications they want to either build on or convert to a microservices architecture. Only roughly 30 percent to 40 percent of applications qualify to be in that first wave of approved projects, and will it take most IT organizations years to complete the first wave of projects.
The primary reason most organizations are embracing microservices is to inject more flexibility into their application environments. But constructing applications as a set of interdependent microservices that can be updated in isolation is not a trivial task. Most IT organizations don’t have an abundance of internal expertise capable of building, deploying, securing and managing these types of applications.
At the same time, however, the pressure on IT to become more agile already has led to widespread adoption of Kubernetes clusters. It’s just that the percentage of application workloads deployed on Kubernetes remains slight compared to other platforms. But as the rate at which microservices-based applications continues to grow, it has become a requirement to provide a standard mechanism to reliably access external storage that isn’t affected by ongoing updates to Kubernetes. There’s a symbiotic relationship between being able to access external data and the number of microservices-based applications that can be deployed. After all, applications that can’t access data beyond what resides in a local Kubernetes cluster won’t be useful.
In fact, most applications an enterprise IT organization builds most likely will be stateful, which means they need access to persistent storage. CSI, in many cases, is likely to be the most widely employed mechanism for achieving that goal within an enterprise data center.