Meta switched to a custom scheduler with eBPF which delivered π 3.28 megawatts of power savings +1.1% on weighted-ads-ranked 28% reduction in service p99 latency 60% reduction in service p99 latency 18% reduction in timeout errors on the critical path engineering.fb.com/2026/07/13/m...
Bill Mulligan πππ
@breakawaybilly.bsky.social
Cloud Native networking, observability, and security with Cilium, eBPF, and Isovalent Newsletter cilium.io/newsletter
"Cillium is like gravity: every time I come across some networking related technology I want to play with, Cillium has an implementation." social.treehouse.systems/@juliancalab...
First stop on the eBPF World Tour was Vienna for IETF and eBPF Vienna Meetup π From someone not coming from a traditional IT background and now be speaking at IETF was a big honor for me. I guess you never know how much your life can change in a decade.
Obviously I like my own quote in this eBPF Foundation press release π βeBPF is becoming the strategic technology of choice for infrastructure platforms and teams. ebpf.foundation/ebpf-foundat...
eBPF Foundation Accelerates Growth with Meta, Toyota Membership, and New Strategic Board Leadership β eBPF
ebpf.foundation
"IMO the maintenance burden of eBPF programs is much much lighter than out-of-tree kernel modules, with the safety guarantee offered by the verifier at the same time."
Actually Maintaining an eBPF Program
Maintaining an eBPF program means surviving multiple kernel versions, architectures, Kconfig options, LLVM changes, and regressions. Here is what I learned while stabilizing tracexecβs eBPF backend.
kxxt.dev
An eBPF-based virtual network dataplane and control plane, written in Rust. Most understanding of eBPF-based networking comes from reading about systems like Cilium rather than building one. Sarena exists to close that gap directly.
GitHub - erwin-kok/sarena: An eBPF-based virtual network dataplane and control plane, written in Rust.
An eBPF-based virtual network dataplane and control plane, written in Rust. - erwin-kok/sarena
github.com
I'll be speaking at IETF in Vienna tomorrow. Since RFC 9669: BPF ISA was published in 2024 a lot has happen and I'll be giving the eBPF ecosystem updates. On a personal note, I still remember when I first heard about IETF in my masters. 9 years later, I'm the one giving the speech π€―
From eBPF podcasts to local meetups and conference talks to hands-on coding labs Teodor Podobnik really covered a lot in his Community & Advocacy Fellowship from eBPF Foundation π Look for applications to open again in the fall π
Community & Advocacy Fellowship β Teodor Podobnik β eBPF
May 2026 marked the end of my inaugural Community & Advocacy Fellowship with the eBPF Foundation, awarded in recognition of my ongoing contributions to the eBPF community.
ebpf.foundation
Cloudflare's case study with the eBPF Foundation is just the teaser for the 8 part blog series we have coming with them. It will cover why Cloudflare bet on eBPF as a platform rather than a collection of point solutions Coming soon β’οΈ
Cloudflare Uses eBPF to PowerSecurity, Networking, andObservability at Global Scale
ebpf.foundation
Looks like hello-ebpf, the Java library for eBPF just got a new homepage and docs β π
hello-ebpf
hello-ebpf is the first and only Java library for eBPF that lets you write both the kernel-side BPF program and the user-space control code entirely in Java β no C files, no Makefiles, no separateβ¦
parttimenerd.github.io
eBPF Foundation is launching the eBPF Company Landscape and we need your help to fill it! Our goal is to track all of the companies leveraging eBPF in their products to show how widely it is used and help end users understand what their vendor choices are π ebpf.foundation/unveiling-th...
Unveiling the eBPF Company Landscape: Mapping the Businesses Powering the eBPF Ecosystem β eBPF
ebpf.foundation
The next eBPF acquisition is in the books with Attribute - FinOps Without Tagging joining DoiT π I see this one as the turning point as it is the first company in the eBPF space to be acquired outside the traditional categories of networking, observability, and security.
Nicolas and team are back at it again. Let them know what is bogging down your cluster networking and help us settle the bet if Gateway API, VMs, or AI use cases will be the most important in the survey this year ποΈ
Help Us Map The State of Kubernetes Networking in 2026
Share how you run Kubernetes networking in 2026. Take the 10-min survey and enter to win prizes!
isovalent.com
Looking at Cong Wang's analysis of the eBPF ecosystem, everything from lines of code to projects on Github to mailing list activity to kfuncs to academic publications and citations is up and to the right π Twelve years in and I feel like we are just getting started
eBPF Trend Analysis
The evolution of eBPF (extended Berkeley Packet Filter) has been one of the most significant developments in Linux system observability and networking over the past decade. As we look at theβ¦
wangcong.org
βThe biggest value for us is that Cilium simply works, and works well,β said Pons. βWith our previous CNIs, failures were painful and hard to resolve. Now we know that once Cilium is deployed, it will work. That reliability is genuinely valuable.β - From Michelin
Michelin
In 2024, Michelinβs Container as a Service (CaaS) platform team underwent a full rebuild of the companyβs Kubernetes platform using open source software, replacing a vendor-based solution that hadβ¦
cncf.io
From AI/GPU neoclouds to hyperscalers to sovereign European clouds and major Kubernetes distributions, everyone is on Cilium Which of our collectable cards in your favorite and should we make a sticker book for next KubeCon?
eBPF-based service dependency mapper for Kubernetes. Builds an accurate topology even through Envoy edge proxies by resolving HTTP Host headers instead of proxy IPs β no sidecars, no instrumentation. Written in Rust, queryable via MCP From Yash Nandwana
GitHub - YashNandwana/qubit: eBPF-based service dependency mapper for Kubernetes. Builds an accurate topology even through Envoy edge proxies by resolving HTTP Host headers instead of proxy IPs β no sidecars, no instrumentation. Written in Rust, queryable via MCP.
eBPF-based service dependency mapper for Kubernetes. Builds an accurate topology even through Envoy edge proxies by resolving HTTP Host headers instead of proxy IPs β no sidecars, no instrumentatio...
github.com
Tom Gillis is always talking about how Cisco is turning into a platform company. You buy into the networking and now you get the security along side it too. Where I'm seeing that on the Isovalent side is turning our eBPF kernel engineering into enterprise workflows through Cisco Cloud Control.
The last 4.5 years with Isovalent have been a ride on a rocket ship π but my favorite ride has always been on the bicycle π΅ββοΈ Starting in August, I'll be going on a semi-batical (yes, Isovalent is cool like that) with my wife by bike around the world.
Write eBPF programs in Ruby π A partition + codegen layer over matz/spinel that compiles eBPF-eligible methods to .bpf.c and keeps the rest native C, with transparent dispatch.
GitHub - yuskesh/spinel-ebpf: Write eBPF programs in Ruby: a partition + codegen layer over matz/spinel that compiles eBPF-eligible methods to .bpf.c and keeps the rest native C, with transparent dispatch.
Write eBPF programs in Ruby: a partition + codegen layer over matz/spinel that compiles eBPF-eligible methods to .bpf.c and keeps the rest native C, with transparent dispatch. - yuskesh/spinel-ebpf
github.com
"eBPF can unlock new pathways towards efficient power management in modern server systems" As we talk about compute in terms of Gigawatt datacenter, this is just one more way that eBPF is helping with the most pressing demands we are placing on our technology stack
Research Update: Managing Server Power with eBPF (Part 2) β eBPF
Contributed By Daniel Wong, University of California, Riverside
ebpf.foundation
A userspace eBPF VM written in Go gBPF tries to ease the validation and testing of eBPF programs by providing control over the input and output memory as well as mimicking the different helpers leveraged when writing eBPF programs.
GitHub - scitags/gbpf: A userspace eBPF VM written in Go
A userspace eBPF VM written in Go. Contribute to scitags/gbpf development by creating an account on GitHub.
github.com
Isovalent is a company of kernel engineers, not CSS majors. When I thought about how investors valued us, I would say it was mainly based on our kernel expertise with enterprise workflows starting to be built in I've put X marks on where I think other startups in the eBPF space land, any guesses?
eBPF based companies capture value by building a technical moat in the kernel and delivering enterprise workflows in user space. When evaluating a pitch deck claiming eBPF capabilities, carefully consider how the founding team is capturing customer value in both kernel space and user space.
Of the five key take away at the top of this conversation from Daniel Finneran and Olimpiu Pop this is my favorite: "While eBPF is famous for networking (Cilium), its true power lies in observability and enforcement across any syscall.
How eBPF Empowers Developers to Observe Inside the Linux Kernel in a Safe and Unintrusive Way
Dan Fineran explores how eBPF has evolved far beyond its roots in packet filtering into a robust, safe way to extend the Linux kernel. He explains how the eBPF "verifier", the security guardrail,β¦
infoq.com
eBPF startup exits have gone through 3 distinct waves π The first wave of acquisitions consisted of performance based observability point solution tools that gave the acquiring company feature or sensor upgrades.
The eBPF Foundation Launches the Investorβs Due Diligence Guide: Why eBPF is Infrastructureβs Ultimate Technical and Economic Moat β eBPF
ebpf.foundation
I've long thought eBPF would be critical for security in the AI era. Paul Arah did a great job capturing how it will help both sandbox AI workloads and defend against AI assisted attacks π
Taming the AI Double Threat with Isovalent Runtime Security
This blog post explores how AI-assisted attacks are changing the assumptions we make about the adversary,
isovalent.com
With Meta running an average of 180 eBPF programs per server, ensuring that they don't conflict with one another is critical to avoid This tutorial is the deep dive into what Meta has learned building and running safe multi-program eBPF environments
BPFChain: Building Safe Multi-Program eBPF Environments β Tutorial | ACM SIGCOMM 2026
Multi-program eBPF deployments are now the norm in production infrastructure β yet the educational landscape has not kept pace. While eBPF fundamentals are thoroughly covered in existing courses andβ¦
conferences.sigcomm.org
"The companies that own that kernel layer and build upward into analytics, policy management, and workflow integration are structurally positioned to displace incumbents who cannot replicate kernel-level access through any amount of engineering effort."
eBPF Replatforming Thesis
ebpf.foundation