# NUMA Published articles for NUMA. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Kubernetes v1.37: Pod-Level Resource Managers graduated to Beta DevFeed: [Kubernetes v1.37: Pod-Level Resource Managers graduated to Beta]() Original publisher: [Read original article]() Author: Kevin Torres Martinez Published: 2026-09-15T18:30:00Z Content type: release Language: en Sources: [Kubernetes Blog]() Topics: [Kubernetes](), [cpu](), [Hardware](), [Latency](), [API](), [gRPC](), [Monitoring]() Tags: [api](), [cpu](), [hardware](), [kubernetes](), [latency](), [monitoring](), [numa](), [release](), [v1]() ### AI overview Kubernetes v1.37 graduates Pod-Level Resource Managers to Beta, disabled by default. The feature lets Kubelet resource managers use pod-level declarations for hardware placement, enabling NUMA-aligned exclusive resources for primary containers while placing sidecars in a shared pod-isolated pool. The release also adds pod-level reporting to the PodResources gRPC API. ### Source excerpt With the release of Kubernetes v1.37, the Pod-Level Resource Managers feature has graduated to Beta status (disabled by default)! First introduced as an Alpha feature in Kubernetes v1.36, this enhancement builds on Pod-Level Resources by equipping Kubelet's Topology Manager, CPU Manager, and Memory Manager to use Pod-level resource declarations (.spec.resources) directly when making hardware placement decisions. Bringing pod-level resources to node managers Before this feature, obtaining exclusive NUMA-aligned CPU cores or memory for latency-critical applications forced cluster operators into an all-or-nothing choice: assign integer resource requests to every container in the Pod, or forfeit exclusive NUMA alignment entirely. For modern workloads running lightweight sidecars (such as logging agents or telemetry exporters), allocating dedicated physical cores to auxiliary containers was wasteful. Pod-Level Resource Managers solves this challenge by enabling hybrid allocation models. The Kubelet can reserve exclusive NUMA-aligned resources for primary application containers while placing non-Guaranteed sidecars into a pod-isolated shared pool. This ensures primary workloads get unthrottled, NUMA-local performance while sidecars benefit from running in a pod-isolated shared pool, enjoying local NUMA alignment and protection from external node interference without consuming dedicated physical cores. What's new in Beta Graduating to Beta brings key operational and API enhancements: Graduation to Beta: Controlled by the PodLevelResourceManagers feature gate, available to opt in (disabled by default) in Kubernetes v1.37. PodResources API Reporting: The v1 PodResources gRPC service (PodResourcesLister) introduces top-level cpu_ids and memory fields on PodResources responses. Monitoring tools and device plugins can query pod-level exclusive assignments directly without double-counting container allocations. Getting started and providing feedback For a deep dive into the tech ## XCP-ng 8.3 LTS: Refreshed Installation ISOs DevFeed: [XCP-ng 8.3 LTS: Refreshed Installation ISOs]() Original publisher: [Read original article]() Author: Gaël Duperrey Published: 2026-08-14T08:09:13Z Content type: release Language: en Sources: [XCP-ng Blog]() Topics: [Server](), [Hardware](), [networking](), [Disk image]() Tags: [amd](), [drivers](), [hardware](), [improvements](), [installation](), [iso](), [linux](), [lts](), [networking](), [numa](), [performance](), [release](), [security](), [server](), [storage](), [update](), [updates]() ### AI overview XCP-ng 8.3 LTS has refreshed installation ISOs containing updates released since the previous installer, installer improvements, enhanced hardware support, and other fixes. The article notes that existing up-to-date hosts already include these changes. ### Source excerpt Refreshed ISOs for XCP-ng 8.3 LTS featuring the latest security updates, improvements, and QCOW2 support. ## Scheduling In Go : Part I - OS Scheduler DevFeed: [Scheduling In Go : Part I - OS Scheduler]() Original publisher: [Read original article]() Published: 2025-03-04T00:00:00Z Content type: tutorial Language: en Sources: [William Kennedy]() Topics: [Go Language](), [Operating system](), [Concurrency](), [Kernel]() Tags: [ardan-labs](), [blog](), [concurrency](), [cpu](), [engineering](), [fundamentals](), [go](), [go-programming](), [golang](), [hardware](), [numa](), [os](), [programming]() ### AI overview This first article in a three-part series explains operating-system scheduling as a foundation for understanding Go's scheduler. It discusses how schedulers interact with multithreaded Go programs and introduces hardware considerations such as processors, CPU caches, and NUMA. ### Source excerpt Although this blogpost was originally published in 2018, the concepts and principles discussed remain crucial for building efficient and performant multithreaded applications in Go now in 2025 (Go 1.24.0). As the Go ecosystem continues to evolve, understanding how the Go scheduler interacts with the operating system scheduler is more important than ever. This three-part series provides a comprehensive overview of the mechanics and semantics behind Go's scheduling, starting with the fundamentals of the operating system scheduler. ## PostgreSQL Hacking Workshop - January 2025 DevFeed: [PostgreSQL Hacking Workshop - January 2025]() Original publisher: [Read original article]() Author: Robert Haas (noreply@blogger.com) Published: 2024-12-12T15:19:00Z Content type: news Language: en Sources: [Robert Haas]() Topics: [PostgreSQL](), [Hacking](), [NUMA](), [Discord]() Tags: [2025](), [discord](), [hacking](), [mentoring](), [numa](), [postgresql](), [workshop]() ### AI overview The article announces January 2025 discussion workshops about Andres Freund's "NUMA vs PostgreSQL" talk from PGConf.EU 2024. Sessions are planned for 1400 and 2000 UTC, with a possible third session at 1700 UTC if demand is sufficient. It also previews PostgreSQL discussions planned for February and March and promotes the PostgreSQL Hacker Mentoring Discord server. ### Source excerpt Next month, I'l be hosting 2 or 3 discussions of Andres Freund's talk, NUMA vs PostgreSQL, given at PGConf.EU 2024. You can sign up using this form. I anticipate that both Andres and I will be present for the discussions, and I'd like to thank Andres and all of the other presenters who have made time to join the discussions and answer questions for their time (so far: Melanie Plageman, Thomas Munro, Andrey Borodin). It has been absolutely great having them join the workshops. Read more " ## ePouta Flavour Halloween 2018 Communiqué DevFeed: [ePouta Flavour Halloween 2018 Communiqué]() Original publisher: [Read original article]() Author: Unknown (noreply@blogger.com) Published: 2018-10-19T10:51:00Z Content type: release Language: en Sources: [CSC - IT Center For Science - Cloud Team]() Topics: [GPU](), [cpu](), [Nvidia]() Tags: [applications](), [cpu](), [deprecated](), [epouta](), [espoo](), [flavors](), [gpu](), [hpc](), [newton](), [numa](), [nvidia](), [openstack](), [performance](), [pike](), [pouta](), [standard](), [westmere]() ### AI overview The communiqué announces changes to ePouta flavors, including public availability of the gpu.2.1gpu flavor with an NVIDIA Tesla V100 and NUMA-aware CPU and memory behavior. It also announces the deprecation of hpc.*.westmere flavors and recommends replacement flavors for new instances. ### Source excerpt We are making some changes to our flavor offering . Here's a summary of the changes: We now have even more GPUs in ePouta. The gpu.2.1gpu flavor is now publicly available and each has one NVIDIA Tesla V100 via PCI pass through. This flavor was already added in July but could only be accessed after requesting it from servicedesk. Now we are making them publicly available. The gpu.2.1gpu flavor is our first NUMA aware flavor. This should speed up CPU & memory heavy applications. Our current version (Newton) of OpenStack is not able to guarantee PCI devices from the same NUMA node too, this is possible to come when we get to Pike or later. Deprecation of hpc.*.westmere flavors This does not affect running instances but in the future you might see changes in the performance and therefor we suggest that you resize your flavor to a newer flavor type. You won't be able to launch new instances with the deprecated flavors. Flavors that will be deprecated and suggested replacement: hpc.medium.westmere - standard.xlarge hpc.large.westmere - hpc.4.10core hpc.xlarge.westmere - hpc.4.10core or hpc.4.20core hpc.largemem.westmere - hpc.4.20core or hpc.3.28core More information about the different flavors can be found here: https://research.csc.fi/pouta-flavours No known Copyright Please acknowledge 'Sir George Grey Special Collections, Auckland Libraries, 1-W682' when re-using this image. ## Diagnosing Unexpected Software Performance Problems on Linux Servers DevFeed: [Diagnosing Unexpected Software Performance Problems on Linux Servers]() Original publisher: [Read original article]() Author: Andy Myers Published: 2017-04-20T00:00:00Z Content type: tutorial Language: en Sources: [Jane Street]() Topics: [Optimization](), [Linux](), [systems](), [Processes](), [Software](), [Process](), [Server](), [Hardware](), [Latency](), [Cache](), [cpu]() Tags: [hardware](), [kernel](), [latency](), [linux](), [linux-kernel](), [numa](), [optimization](), [performance](), [processes](), [processor](), [server](), [systems]() ### AI overview A practical guide to diagnosing unexpectedly poor application performance on Linux servers. It covers verifying the deployed binary and libraries, process CPU affinity, processor cache locality, NUMA placement, hardware topology, and interference from other processes. ### Source excerpt It's often surprising just how much software performance depends on how the software is deployed. All the time and effort you've invested in optimization can be erased by a few bad decisions in scheduler policy, affinity, or background workload on a server.