Software Engineer - Intern - (France)

  • Location:
    Paris, France
  • Area of Interest
    Engineer - Software
  • Job Type
    Intern
  • Technology Interest
    Cloud and Data Center, Networking, Security, Software Development, Testing
  • Job Id
    1399120
Start date: April-June 2023
Duration: 6 months
Location: Hybrid, Paris office & home office

TEAM OVERVIEW

The intern will be part of the Cloud Native Data Plane (CNDP) team within the Security Business Group (SBG) at Cisco. This team contributes to many projects across SBG by developing, maintaining, and customizing the Vector Packet Processor (VPP), which is an open source piece of software part of the fd.io project. VPP is recognized as the most performant software-based packet forwarder available. This makes the CNDP team the ideal place to learn about high-performance data-plane programming.


TOPIC OVERVIEW

The Secure Access Service Edge (SASE) initiative consists of providing end-users with a combination of Software Defined WAN (SD-WAN) services and Network security services, all running in the cloud. In this context, SASE-providers must terminate a very large number of tunnels of various types (e.g., IPsec) in their own infrastructure. Unless the infrastructure is heavily overprovisioned with regards to the aggregated throughput of all the tunnels, Input/Output (I/O) and Central Processing Unit (CPU) resource contention is inevitable at the ingress of this infrastructure. Such contention might lead to unfairness, and poor per-customer latency and throughput guarantees, thus the SASE provider’s inability to honor any kind of network Service Level Objectives.
As a result, it is necessary to design some traffic policing scheme at the ingress of the system (i.e., at the tunnel termination point), that would perform traffic scheduling (i.e., delaying packets belonging to a flow to reduce its burstiness and impact on the downstream system’s buffers) and/or selective packet dropping, to ensure fairness among all tunnels.
While there are plenty of theoretical algorithms that ensure a high level of fairness between flows (for various definitions of fairness), those very often suffer from scalability issues due to the necessity of tracking the state of the different flows traversing the system. Moreover, even for algorithms whose scalability have been studied from a purely “computer-science” perspective (usually meaning asymptotic complexity for a large number of flows), such an approach is insufficient at the levels of throughput targeted by a high-performance software data plane.
In fact, properties such as cache-friendliness, or code vectorizability are more important when the per-packet CPU-budget for traffic policing is in the order of a hundred cycles. To the best of our knowledge, very little research has been done with this perspective in mind


INTERNSHIP GOALS

This internship main general goal is to study traffic policing algorithms from the aforementioned perspective, in the context of SASE ingress. This can be carried out by means of:
• A review of state-of-the-art traffic policing algorithms that ensure reasonable levels of fairness at a minimal complexity
• Select among this list, those which have properties compatible with a high-performance packet-forwarding engine (review data structure cache locality for example)
• Propose efficient implementations in a software dataplane (either the Linux Kernel, Click, or VPP)
• Propose a quantitative theoretical and experimental analysis of those algorithms going beyond traditional asymptotical analysis
• Release produced code in open source and theoretical and experimental results as an academical publication


EXPECTED SKILLS

C programming
Good understanding of core principles of packet switched networks and queuing networks
Ability to jump into a complex code base and hack into it
Ability to read and understand computer science/computer networking research papers
Academic English reading/writing skills

Message to applicants applying to work in the U.S.:

When available, the salary range posted for this position reflects the projected hiring range for new hire, full-time salaries in U.S. locations, not including equity or benefits. For non-sales roles the hiring ranges reflect base salary only; employees are also eligible to receive annual bonuses. Hiring ranges for sales positions include base and incentive compensation target. Individual pay is determined by the candidate's hiring location and additional factors, including but not limited to skillset, experience, and relevant education, certifications, or training. Applicants may not be eligible for the full salary range based on their U.S. hiring location. The recruiter can share more details about compensation for the role in your location during the hiring process.

U.S. employees have access to quality medical, dental and vision insurance, a 401(k) plan with a Cisco matching contribution, short and long-term disability coverage, basic life insurance and numerous wellbeing offerings. Employees receive up to twelve paid holidays per calendar year, which includes one floating holiday, plus a day off for their birthday. Employees accrue up to 20 days of Paid Time Off (PTO) each year and have access to paid time away to deal with critical or emergency issues without tapping into their PTO. We offer additional paid time to volunteer and give back to the community. Employees are also able to purchase company stock through our Employee Stock Purchase Program.

Employees on sales plans earn performance-based incentive pay on top of their base salary, which is split between quota and non-quota components. For quota-based incentive pay, Cisco pays at the standard rate of 1% of incentive target for each 1% revenue attainment against the quota up to 100%. Once performance exceeds 100% quota attainment, incentive rates may increase up to five times the standard rate with no cap on incentive compensation. For non-quota-based sales performance elements such as strategic sales objectives, Cisco may pay up to 125% of target. Cisco sales plans do not have a minimum threshold of performance for sales incentive compensation to be paid.

Share