One of the assumption most IT vendors make today is that containers mainly will be deployed using the same IT infrastructure that historically has been used to support virtual machines. But for a few months now Diamanti has been making the case for a converged appliance that has been built from the ground up to run containers.
At the recent LinuxCon 2016 conference, Diamanti revealed that argument is starting to gain a certain amount of traction with software vendors. MemSQL, a provider of an in-memory database optimized for analytics, became the first software vendor to join the Diametes beta program.
Specifically, MemSQL is employing the Diamanti appliance to accelerate an application development process that makes extensive use of containers to automate the building of custom tests.
Mark Balch, vice president of Products and Marketing for Diamanti, says that before too long IT organizations are going to discover they will need to deploy IT systems natively designed to address the networking and storage requirements of containers in a way that is highly automated. The simple fact is there will be too many microservices based on containers that require different levels of prioritization to manage manually.
Diamanti accomplishes that goal via extensions to the open-source Kubernetes container management framework. Specifically, Diamanti’s FlexVolume contribution automates IO configuration based on user-defined requirements. Diamanti’s scheduler contribution enables the Kubernetes scheduler to factor storage and networking requirements when placing workloads, leveraging a declarative model for developers and container administrators.
The degree to which an upstart such as Diamanti can usurp established providers of hyperconverged platforms remains to be seen. Dell Technologies, Cisco Systems, Nutanix and Hewlett Packard Enterprise (HPE), among others, are already locked in a titanic struggle for control over what is now the fastest growing segment of the data center market.
A big part of the Diamanti strategy assumes that an ever-increasing percentage of containers are going to be deployed on bare-metal servers. Today, the vast majority of containers are deployed on virtual machines or in a platform-as-a-service (PaaS) environment in the cloud. Deploying containers on top of these environments makes it simpler for IT operations teams to manage them. But virtual machines also add overhead to an IT infrastructure environment that is then eliminated when containers are employed as lightweight alternative to virtual machines running on a bare metal server.
Working in Diamanti’s favor are initiatives such as an effort to port OpenStack to Kubernetes being led by Mirantis, Intel and Google. Less helpful to Diamanti are efforts by VMware to make it possible to run containers on top of its hypervisors.
Balch says Diamanti is betting that in many organizations, developers soon will force the issue. IT operations teams may resist the idea of having to embrace a new IT infrastructure vendor. But once developers start noticing the difference between how applications perform on IT infrastructure optimized for containers versus systems optimized for virtual machines, Balch says newer approaches to hyperconvergence eventually will carry the day.