What’s the point: Istio governance, CircleCI, Algorithmia, GraalVM, and KubeOne

What’s the point: Istio governance, CircleCI, Algorithmia, GraalVM, and KubeOne

The Istio project has made some changes to its governance model, giving the community more say in the admin side of things while offering heavily contributing companies a seat at the table as well. 

With a new charter in place, the steering committee responsible for the service mesh will be made up of 13 seats. Nine of those will – proportionally to their contributions – go to members of the top-contributing companies, which right now are Istio instigators Google and IBM/Red Hat, as well as Salesforce. 

The rest are reserved for 4 members of organisations that aren’t represented in the contribution seats just mentioned. Who will end up in a so-called community seat is decided in an election process which has just started. Any project member who doesn’t work for any of the companies with a contribution seat is eligible for candidacy, with nominations open until 6 September. 

Since the announcement states that the goal of the process is “corporate diversification” ensuring “representation from at least 7 different organisations”, we’ll have to wait and see how the project will handle a result that sees more than one person from a company coming out on top. The election starts on 14 September, giving any project member who had their pull request merged in the past 12 months two weeks to cast a vote.

Community seats will be assigned for the duration of a year. In that time, the elected members will help the committee to guide decisions regarding project governance, marketing, and “community health”.

CircleCI tries a new hub for better clarity

CI/CD platform provider CircleCI has decided to bundle its resources under a new user-friendly interface. The so-called CircleCI developer hub is now available via the company’s website, providing a number of jump labels to get engineering teams quicker to relevant documentation, configuration snippets (orbs), Docker images, and APIs. Future iterations hope to see the hub gain an orb development kit, and a “much more streamlined search experience of CircleCI’s documentation”.

Algorithmia introduces team editions

MLOps platform Algorithmia has reworked its product into a Teams flavour, allowing data science and ML groups to work more collaboratively on their AI initiatives. Algorithmia tries to woo developers by promising the sort of lifecycle management often missed in the machine learning realm, including ways to deploy and service models as API and monitor their performance. 

Teams is supposed to make this offering into more of a group affair, so that users have an easier time sharing components such as algorithms and data needed for a machine learning project in a private workspace, while allowing central billing for better overview and throwing better security and scalability into the mix. Teams comes in a basic and a professional version, with basic following a CPU/GPU consumption-based pricing model. Professional users will have to pay an additional $299 per month, but will get better support and customisation options.

Oracle’s GraalVM hits 20.2 

High-performance runtime GraalVM was recently pushed out in version 20.2, which aimed at reducing the tools resource consumption and added a few optimisations to the project. The release will, for example, free up the memory used by libgraal once compilation goes idle, and display more helpful information should a fatal error occur. Meanwhile the Enterprise edition includes an optimisation for removing redundant read/write, array clone() and accesses to the cloned arrays operations, as well as one for loop unrolling

KubeOne celebrates first major release

Amidst the hustle and bustle of KubeCon EU, Kubermatic’s cluster operation automation project KubeOne reached the big 1.0, signaling the project’s production readiness. A good 15 months after its initial release, feature highlights of version 1.0 are a new apply command for cluster reconciliation, and support for provisioning clusters with static worker nodes to help in bare-metal and edge environments. KubeOne now also allows the deployment of any CNI plugin, and comes with a more mature KubeOneCluster API. Details can be found in the project’s changelog.