OpenStack Summit Berlin 2018: Running Your Cloud
The latest cloud and DevOps news from Berlin.
OpenStack has fast become the open source software of choice for providers to create their cloud infrastructure. If the number of attendees at the well-organized event in Berlin’s Messe (main conference venue) was anything to go by, interest in the platform is set to grow further.
To quantify this interest, The 2018 OpenStack User Survey stated more than 10 million computing cores running OpenStack in over 60 countries. OpenStack is one of the three most active open source projects, alongside the Linux kernel and Chromium, with 70,000 commits in the last year, and an average of 182 changes per day during the OpenStack Rocky cycle.
The keynotes each day focussed on the growth of OpenStack, the status of the community, and new use cases such as AI, Machine Learning, edge computing and GPU networks. The keynotes reflected the wide variety of industries that use OpenStack, with presentations from representatives of banks, textile producers, car manufacturers, and telcos. The OpenStack foundation also announced a pilot program to incubate projects working in these fields.
The announcements of updates, partnerships and product releases were constant. Alongside this were technical deep dives into OpenStack features, integrations and extensions. I’ll distill some of what I saw at the summit, which was a small selection, you can find videos for the session on the summit website.
Infrastructure as Code
announced its own container, VM and bare metal deployment toolchain aimed at Telcos. Despite only having a release candidate available, AT&T are already trialing Airship with some of its 5G rollouts.
Kata Containers announced an interesting idea, creating a new lightweight VM that gives the experience of containers, but with the total isolation that VMs offer. I’d be interested in trying them in the future to see if the benefits are real.
Zuul and Automium were projects that caught my eye by enhancing current CI/CD workflows with running varied integration tests against commits. They do this using Ansible and Terraform scripts to run your commits against multiple platforms and other commonly integrated projects before declaring a build successful.
Alongside a large presence from RedHat and a smaller presence from SuSe, Canonical were out in force at OpenStack Summit, with keynotes from Mark Shuttleworth and multiple demos and talks. Thanks to the predominance of Ubuntu images in cloud providers, Ubuntu is the number 1 platform for OpenStack on public clouds, and at the summit there was a specific focus on their Metal-as-a-Service (MAAS), AI, Multi-Cloud and Telco uses.
While I initially struggled to understand what Kaloom’s Software Defined Fabric (SDF) offered over other options such as Software Defined Networks (SDN), their recent funding round of over $10 million shows that the need to connect applications spanning multiple data centers is big business. Kaloom’s solution is often found at the hardware level inside programmable switches and white boxes, and their new “Virtual Central Office” solution adds a RedHat-based option for data center providers to manage their network(s).
Looking to 2019
For OpenStack users and implementers, “Train” is the name of the next OpenStack release, likely released at the Open Infrastructure Summit (OpenStack summit’s new name). In Q4 2019 the summit will be held in China for the first time, keep an eye on openstack.org/summit for more details.