HPE Partners with Portworx on Bare-Metal Server for Containers
As containers running on top of Kubernetes and other orchestration platforms continue to accelerate, many believe there soon will be a wave of IT infrastructure upgrades driven by I/O requirement of microservices based on containers.
To help make it simpler to make that transition, Hewlett-Packard Enterprise (HPE) and Portworx, a provider of storage software optimized for containers, have combined their respective expertise to create a reference platform for deploying stateful containers applications on top of bare-metal servers.
The reference architecture is intended to make it possible to stand up a bare-metal HPE server running an instance of Kubernetes accessing storage systems managed by Portworx in less than 30 minutes. HPE has already launched an IT operations platform optimized for containers along with forming alliances with providers of clustering platforms from Mesosphere and Docker Inc. that can be deployed along with or as alternatives to Kubernetes.
Eric Han, vice president of product management for Portworx, says the company’s approach to data storage in a container environment can be employed on top of a bare-metal server, as in the case of this reference architecture, or in concert with virtual machines running in an on-premises environment or a public cloud.
While containers today mainly are deployed on top of virtual machines, Han says I/O performance issues will become more commonplace as containers proliferate. That will drive more organizations to embrace bare-metal servers running Kubernetes as an alternative to existing commercial virtualization software. That doesn’t mean virtual machines will go away anytime soon, but it does mean IT organizations will need to support virtual machines and bare-metal servers side by side for a while.
Han says the Portworx value proposition is to provide a common layer of software for managing container storage across a hybrid computing environment. Han contends that while most of the usage to data for containers has centered around stateless applications, there is also a surge of stateful applications under development that will need to access modern storage systems. In addition, Han notes there also will be a wave of existing stateful legacy applications that will be containerized, which will also serve to increase demand for storage systems optimized for containers.
It’s unclear just yet how many containerized applications will be deployed on-premises or in the cloud, or whether those applications will be deployed on virtual machines or the cloud. There’s also a case being made to unify containers and a lighter-weight instance of a hypervisor. But Han says many IT organizations want to be able to pivot in any direction as either their internal culture changes or various technologies that make up the container ecosystem mature. But at the same time, many of them may also be underestimating all the data management challenges inherent in support of microservices-based on containers that are likely to either move or be replaced by another container at any given moment.
