NET-210 · Kernel Networking · Advanced

DPDK & Kernel Bypass — full syllabus

When to leave the kernel network stack entirely, and what you give up when you do.

Duration3 full days in person · 6 half-days online
Cohortmax 14 in person · 20 online
Pricefrom SAR 9,000 in person · local pricing per city
Delivery35% principles · 20% guided investigation · 45% engineering studio

Who this course is for

Engineers evaluating or operating kernel-bypass data planes who need an honest account of what DPDK buys, what it costs, and when XDP or the tuned kernel stack is the better answer.

Prerequisites

Course outline

Day 1 — the bypass architecture

  • What the kernel path costs: context switches, copies, per-packet work
  • Poll-mode drivers and the run-to-completion model
  • UIO and VFIO: taking a NIC away from the kernel
  • Hugepages, memory pools and mbufs
  • lcore assignment and NUMA placement rules

Day 2 — programming the data plane

  • EAL initialisation and device probing
  • rx/tx rings, descriptors and burst processing
  • A minimal forwarder, line by line
  • Multi-queue, RSS and scaling across lcores
  • What happens to tcpdump, ethtool and ss when the kernel no longer sees the NIC

Day 3 — evaluation and operations

  • Benchmarking honestly: DPDK vs XDP vs the tuned kernel stack
  • NUMA and hugepage misconfiguration penalties
  • Operational cost: debugging, monitoring, upgrades, onboarding
  • Failure modes: what a poll-mode loop hides until it melts
  • Decision framework: bypass, XDP, or fix the kernel path

Hands-on labs

  1. Lab: bind a NIC to VFIO, allocate hugepages and reach line rate with testpmd
  2. Lab: write a minimal DPDK forwarder and measure throughput and latency against the kernel path
  3. Lab: move the same workload across NUMA nodes and quantify the placement penalty
  4. Lab: run one filtering task on the kernel stack, XDP and DPDK; chart throughput, CPU and debuggability

Capstone project

Run a controlled bake-off and write the recommendation: one packet-processing workload implemented on the tuned kernel path, XDP and DPDK, with a shared measurement harness — then deliver a decision memo covering throughput, tail latency, CPU cost and operational burden, defended in a review session.

What you leave with

Upcoming dates

DatesWhereSeatsEarly birdRegular
25 Oct – 27 Oct 20263 full days RiyadhIn person · KAFD Conference Centre 10 of 14 —SAR 9,000
1 Nov – 3 Nov 20263 full days Kuwait CityIn person · Al Hamra Tower 5 of 14 —KWD 740
1 Nov – 3 Nov 20263 full days MuscatIn person · Knowledge Oasis Muscat 10 of 14 —OMR 920
8 Nov – 15 Nov 20266 half-days Gulf bandLive online · 09:00–13:00 GMT+3 4 of 20 US$1,580until 9 OctUS$1,750
9 Nov – 11 Nov 20263 full days OttawaIn person · Kanata North Tech Park 5 of 14 CAD 2,930until 10 OctCAD 3,260
16 Nov – 18 Nov 20263 full days TorontoIn person · MaRS Discovery District 10 of 14 CAD 2,930until 17 OctCAD 3,260
16 Nov – 18 Nov 20263 full days LondonIn person · Shoreditch Works 5 of 14 GBP 1,680until 17 OctGBP 1,870
16 Nov – 23 Nov 20266 half-days Europe bandLive online · 09:00–13:00 CET 9 of 20 US$1,580until 17 OctUS$1,750
16 Nov – 23 Nov 20266 half-days Americas bandLive online · 13:00–17:00 ET 14 of 20 US$1,580until 17 OctUS$1,750
23 Nov – 25 Nov 20263 full days BerlinIn person · Factory Görlitzer Park 10 of 14 EUR 1,990until 24 OctEUR 2,210

Book a seat, or bring this course to your team

Seats can be reserved online; private delivery runs on-site or live online, adapted to your stack.

Course page & booking

Questions about fit or prerequisites? Email hello@kernelsystems.academy. To save this syllabus, print this page to PDF from your browser.