 |
VOOZH | about |
|
VOOZH | about |
We’re so glad you’re here. You can expect all the best TNS content to arrive Monday through Friday to keep you on top of the news and at the top of your game.
Check your inbox for a confirmation email where you can adjust your preferences and even join additional groups.
Follow TNS on your favorite social media networks.
Become a TNS follower on LinkedIn.
Check out the latest featured and trending stories while you wait for your first TNS newsletter.
Normally, a software release with a lot of version numbers in it isn’t what’s considered a “major” release. That’s known as a “point” release. The big ones are usually relegated to simple X.0 or XX.0 designations.
Well, don’t be fooled by overlooking Kubernetes’ v1.31.0, which became generally available today. Its release leader, Angelos Kolaitis, says he considers this a “major minor release,” and that it’s important enough to warrant more than cursory attention.
Kubernetes, also known as K8s and originally developed by Google, is an open source platform designed to automate the deployment, scaling and management of containerized applications. It has skyrocketed in popularity and production use in IT systems in the last decade.
This latest edition (Kubernetes churns out three versions a year on a disciplined schedule) represents significant advancements in support of AI/ML workload processing and in networking generally, Kolaitis said.
For support of AI/ML workloads, v1.31.0 introduces a volume type for (OCI) images, which enables developers to easily change a large language model used in a workload by simply changing out its image. OCI images are now used as volume sources in Kubernetes, so changing models or model weights is simplified.
“Developers using one particular model now can just change some OCI elements in the same way that they already know how to change images – for deployments when you want to upgrade or when you want to try something new. This is a very familiar process,” Kolaitis said.
OCI (Open Container Initiative) refers to a set of open standards and specifications that govern the creation, distribution and execution of container images.
The new edition exposes information about hardware devices used by pods, such as GPUs, in a more standardized and efficient way. Finally, it provides initial support for the new Device Resource Assignment (DRA) feature, which helps standardize the process of accessing and managing hardware accelerators, such as GPUs.
In networking, v1.31.0 improves Kube-proxy, a critical network component responsible for service discovery and load balancing within a cluster, with a new nftables bucket, which helps address performance limitations. Buckets are used to implement rate limiting and traffic-shaping mechanisms in nftables. They help prevent network congestion, ensure fair bandwidth allocation and protect against potential attacks, such as denial of service (DoS).
The new edition also continues to streamline and stabilize its core networking components, providing more reliability and robustness without requiring changes from users, Kolaitis said.
The Device Resource Assignment (DRA) feature is an important step forward in standardizing the process of accessing hardware accelerators, Kolaitis said. DRA enables the allocation and management of hardware devices, such as GPUs, field-programmable gate arrays (FPGAs) or network interface cards (NICs), to specific pods much more efficiently.
Key data points about DRA:
Other improvements in v1.31.0 include:
Go here to access the complete blog post detailing the v1.31.0 release.
Go here for the GitHub release notes.
