DevOps – the biggest industry-changer in IT

DevOps – the biggest industry-changer in IT

VMworld is a global twin-conference for virtualization & cloud computing organized by the leading virtualization vendor: VMware. ASP has a long tradition of attending – as VMware’s partner – to the European version of VMworld, this year it was held in Barcelona, Spain on 12-16/10/2015.

As expected VMware presented their newest versions of their products; with better performance, higher scalability, new features, requested enhancements and less latency. This blog is not about all of these items. I’m going to write about the biggest industry changer in today’s IT: DevOps.

So what is DevOps again?

For starters it’s not a product that can be bought, it’s an approach to bring the development team closer to your operational team. It’s about processes and people driven by rapid code deployment. The term “DevOps” became popular through a series of conferences that was first organized in back 2009 in Belgium: DevOpsDays with a main focus agility and rapid code deployment.

Since then, there have been DevOps Days conferences held in many countries worldwide.

VMware views DevOps as a key influencer in today’s IT organization, enabling development teams to deliver applications faster with agility by leveraging containers and automation.

Containers you say?

A container instance is a virtualization method where the kernel of an operating system allows multiple isolated user process namespaces, instead of just one. Such containers may look and feel like a real server from the point of view of its owners, users and applications. Typically each container has a dedicated filesystem but is leveraging a shared OS kernel.

Why do we want containers? Developers are crazy about containers as it offers them portabilityspeed and lightness, unfortunately it doesn’t match the current reality and needs of the Infrastructure (team). Their focus is on: data persistency, rich SLA’s, consistent management, security & network services.

In a reaction to the explosion of interest in the potential of application containers VMware initiated one year and a half ago a new project: Bonneville. The main goal of this project is to provision containers directly to virtual infrastructure as first-class citizens and to do it efficiently, quickly and transparently using the same tools that developers & IT ops are using today. By integrating containers deeply into vSphere they reached a one-to-one mapping which enables developers and Infrastructure guys to share the same abstraction. vSphere integrates tightly with: Docker, Kubernetes (by Google), Hadoop, Picotal CF & MESOS.

But how does it work?

It all starts with a new platform called Photon, which consist of two core components: Photon controller which leverages project Lightwave that takes care of the identity & access management challenges in cloud native apps and Photon OS. Both the Lightwave & Photon OS projects are released as open source software.

The Photon controller and Photon OS have a clustered design and therefore delivers linearly on scale & high-availability, both have the following characteristics:

  • The Photon OS is built on top of ESXi and has a very small footprint (25MB).
  • Every container is an instant clone (<200ms) from this Photon OS base image.
  • Containers are linked by a virtual controller host (VCH) which all run in a single resource pool.
  • Containers communicate between each other and the VCH through an isolated internal network.
  • The VCH converts the container framework control commands into vCenter commands.


The architecture differentiates itself by:

  • Speed: 1000s of workloads in seconds
  • Scale: 100K -> 1M containers
  • Rich API set
  • Distributed management & runtime

It’s safe to conclude that VMware embraces the future by building solutions that incorporate DevOps’ way of working, in short: The future looks promising!

Note that the functionalities, specifications & features which are discussed in this article are currently in private beta testing phase, they can therefore deviate in the final product(s).

Leave a Reply