Juniper Networks has advanced its Juke container storage platform by adding support for both Kubernetes and the associated Container Storage Interface (CSI).
Scott Sneddon, senior director and evangelist for multicloud solutions at Juniper Networks, says the addition marks the first update to the container storage platform since Juniper Networks moved to acquire HTBase late last year.
Version 2.2 of Juke provides the file system required to make it possible for Kubernetes clusters to access distributed persistent storage resources in addition for support for volume snapshots, clone management and simple installation tools.
Sneddon says Juniper Networks now views distributed storage as a complementary extension to its efforts to create a software-defined network (SDN) that is based on its Contrail platform and spans multiple cloud computing environments, which increasingly will standardize on Kubernetes. As Juniper Networks continues to invest in Juke, IT organizations should expect to see Juke being deployed in both on-premises IT environments as well as on public clouds, says Sneddon.
In the future, Juniper Networks envisions Juke playing a role in automated multicloud arbitrage based on changing day-night usage patterns, performance and cost optimizations and the movement of workloads between private and public clouds.
Interest in distributed approaches to accessing persistent storage from within containerized applications is on the rise for two reasons.. For one, enterprise IT organizations are starting to build more stateful applications using containers, as the first generation of containerized applications were largely lighter-weight stateless applications that did not require access to persistent storage. The second major driver is a desire to move existing legacy stateful applications on to a Kubernetes platform. Once that migration is completed, organizations typically will start to modernize those monolithic applications by carving them up into a set of more manageable microservices whenever possible.
Juke enables organizations to deploy those applications to access persistent storage regardless of whether they are deployed on a virtual machine, a bare-metal server or in a public cloud, notes Sneddon.
Juniper Networks has not yet signaled any intention to make Juke available as an open source project, but Sneddon says the company is committed to working on storage-related initiatives led by the Cloud Native Computing Foundation (CNCF), which oversees the development of Kubernetes.
Competition between traditional storage vendors, startups and new entrants such as Juniper Networks is likely to be fierce in the months ahead as the volume of data being attached to Kubernetes clusters achieves critical mass. Less clear for now is whether the decision to go with one storage platform over another will be driven by developers, a DevOps team or storage administrators, most of whom have not yet had much exposure to Kubernetes. In fact, many organizations are adopting hyperconverged infrastructure (HCI) platform in on-premises IT environments that unify compute and storage management under a single administrator.
Regardless of where stateful containerized applications are deployed, they will need access to distributed storage resources via a software-defined platform. The only real question now is determining which of the myriad evolving approaches to achieving that goal to pursue.