Kubernetes is an open-source software for orchestrating the deployment, scaling and management of containers.  It was originally designed and developed by Google but is now an open-source software maintained by the Linux Foundation under the Cloud Native Computing Foundation (CNCF) umbrella.

Containers enable multiple programs to run in packaged virtualised environment; aka containers, where they can only see and use the contents and resources assigned to that container.

This is significantly different from the traditional approach, where programs ran in an operating system (OS) environment and could see all the resources in the system running through that OS – with some filter from user profiling.

So multiple programs (can) run inside one container and multiple containers can run in the same orchestration environment.  You can just run one program in a container as well if you want to.

This enables more efficient use of resources by the programs and also helps avoid access and, potentially, security issues, associated with the more traditional approach; where the whole system was exposed by the OS – depending on user profiling.

Containerised workflows are controlled by Kubernetes as the orchestration software (for example) and can be load-balanced across a large resource pool of workers (compute resources).

In other words, you can take a lot of server computers, make them into a resource pool and fill those resources very efficiently with Kubernetes & Docker. Kubernetes ensures the workloads are scaled for optimal performances across your server pool running the programs your users need.  A lot of that efficiency comes from not having resources ‘ring fenced’ for programs which ‘might need them’, which is a key issue with the traditional approach or a static configuration with other cloud technologies.  When you don’t need the resources they can be powered down or allocated to run other processes or containers and Kubernetes can manage that for you when given the right configuration.

This is one big advantage of running Kubernetes on the Kontron Symkloud architecture, it enables all those compute resources to be fielded in a very dense, power efficient way while also integrating the L2/L3 ‘leaf’ switch into the same architecture.  This means less rack space and power consumption for more output, as well as a far greater level of control over the resources and environment from fewer touch points than would be possible with elements – while still retaining the benefits of separate data-plane and control-plane traffic and traffic-management through the switches.  Lastly, because our switches can be managed by SDN controllers like top of rack switches, the network itself, between the resource elements, can be automated and ‘self-learning’ with a high level of redundancy.

As Kubernetes is open-source, there is no software lock-in by Kontron.  You can equally run Docker on Symkloud over CentOS or Ubuntu operating systems or leverage Kubernetes alongside Docker to orchestrate multiple units.  What we have done is build value by testing and determining the optimal power and compute package to both run the Kubernetes orchestration and control environment and still provide plenty of worker resources to be used by the containers.  This can save hours, days or even weeks, depending on how much experience and resources an operator or service provider has to throw at the issue.

Once it’s up and running, there is a secondary benefit that the Kontron Symkloud can be added to or upgraded.  It’s not a single-use vehicle or dedicated appliance.  Kontron have been shipping the same MS29xx frame and backplane since 2012-2013 when we introduced it, so customers since then have been able to progress through GbE to 10GbE to 40GbE and now 1, 10, 25, 40, 50 or 100Gbps interconnects and through three generations of Intel® processor sleds.