SDN & P4 Offload on DPU Intern

Kalray; Montbonnot-Saint-Martin; Hybrid; EUR 12,000 - 18,000; Full time

Benefits offered by this job

Hybrid work

Meal vouchers

Team building activities

Job summary

Kalray invites you to join its DPU Services internship in Montbonnot (France) to work on P4 programmable networking for SmartNICs/DPUs. You will help evaluate P4 for offloading SDN functions, in a lab with TRex/Spirent traffic generators, contributing to high-performance networking projects.

The internship offers exposure to latest-generation hardware, agile teamwork, and hands-on experience in benchmarking, validation, and architectural recommendations for Kalray’s data-center solutions.

Qualifications

• C networking and Linux development skills.
• C and Python on Linux, Bash, Git.
• P4 programming on hardware targets.
• Familiarity with OVS offload and hardware-based connection tracking.
• Network benchmarking at 100 Gb/s and beyond.

Responsibilities

• Map out P4 use cases in the SDN ecosystem.
• Benchmark P4-based SDN solutions on SmartNICs/DPUs (AMD Pensando, Intel IPU, NVIDIA BlueField).
• Implement a P4-offloaded OVS switch controlled by P4Runtime.
• Implement a stateful L4 firewall offloaded to P4.
• Present benchmark reports and architectural recommendations.

Skills

C networking

Python

Bash scripting

Git

P4 programming

OVS offload

Network benchmarking

Linux

Technical English

Education

Engineering student (3rd/4th year)

Tools

DPDK

eBPF/XDP

Open vSwitch

Job description

As part of our DPU Services team, which is responsible for all custom networking development based on the SDKs of commercially available SmartNICs and DPUs (NVIDIA BlueField/ConnectX, AMD Pensando, Intel IPU), you will actively contribute to projects involving P4 programmable networking. You will help evaluate P4 as a foundation for offloading SDN functions onto these cards, supporting the company’s activities and growth ambitions.

In next-generation datacenters and clouds, DPUs (Data Processing Units) and SmartNICs take care of tasks that server CPUs should no longer have to handle: virtual switching, security, encryption, storage, telemetry, and more, at 200 or 400 Gb/s. The P4 language makes it possible to program their packet-processing pipelines directly, and it has proven itself at very large scale: Microsoft Azure relies on P4 specifications (the open-source SONiC-DASH project) to offload its stateful networking services onto DPUs. But how capable are P4, its tools and its performance in practice across different manufacturers? Your internship will answer this question through measurements and prototypes.

This final-year internship will give you the opportunity to gain first-hand experience in the world of DPUs, work with the latest-generation hardware in our lab (servers, SmartNICs/DPUs, TRex and Spirent traffic generators), develop your high-performance networking skills and put the knowledge acquired during your studies into practice in an innovative and stimulating environment.

As such, your main responsibilities will be the following:

• Map out P4 use cases in the SDN ecosystem: architectures (PSA, PNA), control plane (P4Runtime), open-source projects (SONiC-DASH, OPI, ONOS, P4-DPDK, eBPF backend, etc.) and use cases (virtual switching, VXLAN/Geneve overlays, load balancing, firewalls, INT telemetry, 5G UPF, etc.), assessing their actual maturity;
• Benchmark P4-based SDN solutions on commercially available SmartNICs/DPUs — AMD Pensando (SSDK, P4-16), Intel IPU (programmable FXP pipeline in P4), NVIDIA BlueField-3 (DOCA Pipeline Language, derived from P4-16) — and compare them with native offload on ConnectX-7/BlueField-3 (ASAP², OVS-DOCA), assessing language coverage, compilation and debugging toolchains, P4Runtime integration, performance and ecosystem sustainability;
• Implement a P4-offloaded OVS switch on the selected platform: a P4-defined L2/L3 and VXLAN switching fast path, controlled by the Open vSwitch control plane via P4Runtime, then benchmarked against native offload in terms of throughput, latency and host CPU load;
• Implement a stateful L4 firewall offloaded to P4: track TCP/UDP connections directly in the data plane (PNA architecture “add-on-miss” tables, flow aging), with filtering policies managed by the controller, then characterize performance in terms of connections per second, concurrent connections, throughput and resilience to SYN floods;
• Present and showcase your work through a benchmark report, architectural recommendations for the team’s solutions, demonstrators and, where appropriate, open-source contributions.The internship will result in the development of an integrated prototype covering a functional scope defined at the beginning of the internship, together with tests, documentation and an analysis of limitations and possible next steps.

You will be supervised by DPU Services architects and engineers and will work using agile methods within a rapidly growing team, where your results will directly influence our technical decisions.

REQUIRED AND DEVELOPABLE SKILLS

• C Networking: Ethernet, IP, TCP/UDP, L2/L3 switching, VLAN/VXLAN, principles of filtering and connection tracking;
• Development: C and Python on Linux, Bash scripting, Git;
• Methodology: Enjoy experimenting and taking measurements, rigor in test protocols, critical thinking when analyzing result.
• P4 programming on hardware targets,
• OVS offload and hardware-based connection tracking,
• network benchmarking at 100 Gb/s and beyond.
• Some initial exposure to P4 (university project, p4lang tutorials with BMv2/Mininet), DPDK, eBPF/XDP or Open vSwitch would be appreciated.

PROFILE

• Third-Third-year engineering student specializing in networking, computer science or embedded systems;
• Curious, ambitious and passionate about high-tech and applications based on new technologies;
• Enjoy working as part of a team and taking initiatives;
• Enthusiastic about being part of a human adventure and working in an agile environment;
• Comfortable in English, particularly in a technical context.

Are you passionate about high-tech and cutting-edge technologies? Do you want to be part of a human adventure and develop in an agile, dynamic and collaborative environment? This internship is for you.

CONTRACT

• Contract type: Internship agreement
• Duration: 6 months
• Compensation: Depending on level of study
• Collective agreement: Metallurgy
• Benefits: Hybrid work, meal vouchers, team building and other activities such as laser tag, escape games, sailing, skiing, go-karting, karaoke and much more!
• Location: Montbonnot(38)

ABOUT KALRAY

Founded in 2008 as a spin-off from CEA, Kalray is a leading European semiconductor company, backed by major industrial investors such as Safran, NXP Semiconductors and Bpifrance. With more than 30 patent families and more than €100 million invested in R&D, our technology is paving the way for a new era in data processing.

Thanks to their patented “manycore” architecture, Kalray’s MPPA intelligent processors (known as DPU or xPU) are capable of managing multiple data streams in parallel – and without bottlenecks – enabling applications that process massive amounts of data to become smarter, more efficient and more energy-efficient, complementing traditional approaches (CPU and GPU).

Kalray’s offering includes hardware and software acceleration solutions, as well as a co-development – or design-ondemand – offering for processors and acceleration solutions optimized for specific uses in high-growth sectors such as data centers, AI gigafactories, 5G networks, Industry 4.0, space and defense, and many others.

Above all, Kalray is a team of passionate, enthusiastic and committed people who place excellence and collaboration at the heart of their work. Whether you are an intern or an apprentice, we will do everything we can to help you fully express your talent, develop your potential and contribute to the company’s success.

Our offices are located near Grenoble, in the French Alps, and we also have additional offices near Nice, in Sophia Antipolis, as well as in Romania.