Past Events

Explore the past DPDK Summit sessions

DPDK Summit Userspace 2016 - October 20th - 21st, 2016

Userspace 2016 was a developer forum which focused on the elements of DPDK which are most pertinent to the open source software community members. The two-day event at the Clayton Hotel in Dublin included highly interactive discussions on the latest features and upcoming changes to DPDK.

Bruce Richardson

Intel

DPDK Roadmap

This session will be an open discussion on the DPDK roadmap. It will cover topics such as:
- What do people plan to contribute to 17.02/17.05?
- What gaps exist in the roadmap?

Stephen Hemminger

Microsoft

DPDK Hyper-V Support

This presentation will cover the strategy and implications of Hyper-V support in DPDK.

Jasvinder Singh

Intel

Cristian Dumitrescu

Intel

NFV Use-case Enablement on DPDK and FD.io

This presentation will discuss the rapid development of NFV use cases such as a virtualised provider edge router (vPE) using the DPDK and VPP framework on Intel multicore CPUs.  Additionally, this talk will focus on enabling DPDK Hierarchical Quality of Service Schedular (HQoS) in VPP framework. This will include the configuration of HQoS instances and their execution in VPP.

Olivier Matz

6WIND

Mbuf Changes

This session will be an open discussion on mbuf changes that are required, including mbuf extension for external data.

Declan Doherty

Intel

Damjan Marion

Cisco

Sergio Gonzalez Monroy

Intel

Cryptodev and IPsec Acceleration

This presentation will consist of 3 parts:

  • A comprehensive overview of the cryptodev framework in DPDK, including its architecture, poll mode drivers, crypto application development, and details on the future roadmap.
  • An overview of the Vector Packet Processing (VPP) project in FD.io.
  • A description of the integration of cryptodev into VPP to accelerate IPsec.

Keith Wiles

Intel

Transport Layer Development Kit (TLDK)

This presentation describes the Transport Layer Development Kit (TLDK) project in FD.io. It will include some performance measurements of TLDK.

Prof KyoungSoo Park

KAIST

mOS Networking Stack: A Specialized Network Programming Library for Stateful Middleboxes

Stateful middleboxes such as intrusion detection systems and stateful firewalls rely on TCP flow management to keep track of on-going network connections. Implementing complex TCP state management modules for network appliances in high-speed networks is difficult. This is especially more challenging due to the lack of a reusable networking stack that provides a development interface that monitors fine-grained flow states for stateful middleboxes. This presentation describes the middleware Operating System (mOS) which aims to address this gap.

Hanoch Haim

Cisco

TRex Traffic Generator – Stateless Support

This presentation will give high level of the capability/design of the new stateless features in Trex and how it is used by the FD.io project.

Tim O'Driscoll

Intel

DPDK Project Growth

As DPDK continues to expand, we need to consider what changes need to be made in order to facilitate that growth. One obvious change would be to move the project to an independent body such as the Linux Foundation.

Thomas Monjalon

6WIND

DPDK optimal performance everywhere
(Rallying with a Formula 1)

This presentation will be about the fundamentals of DPDK. First of all, the DPDK userspace drivers process packets at lightning speed. And it runs on several architectures with a long list of supported devices. But the real uniqueness of DPDK is to combine optimal performance with a broad hardware support and à la carte packaging. Surprisingly, there are always some new promising optimizations to unlock the full power of the hardware. Also, after 5 years of existence, the project is still making some progress to be easier to use in more and more environments. Finally it is essential to remember that these improvements are possible thanks to an amazing and fast growing Open Source community.

Bruce Richardson

Intel

Identifying and Fixing Performance Bottlenecks

This presentation will focus on identifying and fixing performance bottlenecks using the Intel® VTune™ Amplifier. An example from the i40e driver will be used to illustrate the process.

Shreyansh Jain

NXP

Re-structuring the DPDK Device Driver Framework

Expanding DPDK to support non-PCI devices e.g. platform bus devices in a SoC

Rony Efraim

Mellanox

Mellanox Bifurcated DPDK PMD

Mellanox PMD is based on Bifurcated driver and allows the kernel (netdev) and more than one PMD to run on the same PCI. If the DPDK app is not setting a rule to steer this traffic it will be processed by the kernel.

Jingjing Wu

Intel

Intel I40E Bifurcated Driver

DPDK is known to build the high performing data plane workload on Intel Architecture and platform, a real world packet processing workload often relies on heavily on the Linux kernel and its large stack for the control plane design and implementation. As a known limit, Linux performance is not sufficient for high speed data plane workloads. In order to combine the advantages of both, a few key technical components can be used to make DPDK and kernel stack work and evolve independently.

  • A high speed data path between Linux kernel and DPDK. (Kernel NIC Interface, KNI, TAP).
  • A high speed data traffic direction into Linux Kernel and DPDK. (Bifurcated driver, Virtualization)

Ferruh Yigit

Intel

Interworking with the Linux Kernel

This session will be an open discussion on the challenges of interworking with the kernel, the inability to upstream enhancements like KDP/KCP to either the kernel or DPDK, and possible next steps.

Emma Foley

Intel

Maryam Tahhan

Intel

Daniel Parker

Intel

Monitoring your DPDK Application/Interfaces with Collectd

This presentation describes how to leverage the DPDK xstats API to monitor your interfaces with collectd (a system statistics collection daemon).

Christian Ehrhardt

Canonical

Luca Boccassi

Brocade

deb_dpdk - Challenges and Opportunities when Packaging DPDK

DPDK is special in many ways which make packaging and distributing it a more interesting effort compared to many other programs or libraries. This talk is about challenges we faced, what we can learn from them and where we expect potential areas of improvement in the future. Working together on those should help DPDK distribution and growing a higher adoption rate.

John McNamara

Intel

Christian Ehrhardt

Canonical

Luca Bocassi

Brocade

Usability (including packaging, stable releases, LTS releases etc.)

This session will be an open discussion on usability, including topics such as packaging, stable releases, LTS releases etc.

Qian Xu

Intel

Testing and Continuous Integration

Presentation from Qian on DPDK CI Enhancements.

This presentation will focus on testing, performance testing and CI, and how these can be further improved.

Rony Efraim

Mellanox

Open vSwitch DPDK Acceleration Using HW Classification

How to use HW classification to accelerate OpenvSwitch DPDK. NIC HW can classify the packets and return flow id that can be used instead of classification the packet by software. NIC HW can classify the packets drop and count them in order to preform aging and statistics.

Cian Ferriter

Intel

An Implementation of a P4 Software Dataplane Using Open vSwitch

P4 is an emerging standard for programming dataplanes. Although its initial applicability is for hardware dataplanes (in particular switching ASICS), its paradigm shifting approach to programming the dataplane can equally be applied to software dataplanes with interesting consequences. The first mainstream implementation of a software P4 dataplane is likely to be via Open vSwitch and is based on DPDK. This presentation will talk about this work and what it means to the DPDK community.

Ian Stokes

Intel

Sugesh Chandran

Intel

Bridging the gap between hardware functionality in DPDK applications and vendor neutrality in the open source community

To date there has been multiple efforts to make use of hardware features and functionality in the fast-path of OVS to improve performance e.g. Intel XL710 VxLAN tunnel optimization using flow director feature, Intel XL710/82599  packet type identification etc. However implementations to date have been hardware specific and as such are not acceptable to the Open vSwitch community due vendor neutrality. Feedback from the Open vSwitch community indicated that what is required is a common hardware API that is vendor neutral and easily consumable.

Paul Emmerich

Technical University of Munich

Testing and Benchmarking Applications with MoonGen

MoonGen is a fully scriptable high-speed packet generator built on DPDK and LuaJIT. It can saturate a 10 Gbit/s connection with 64 byte packets on a single CPU core while executing user-provided Lua scripts for each packet. Multi-core support allows for even higher rates, we have evaluated it at rates above 100 Gbit/s at 200 Mpps.

Jim St. Leger

Intel

Closing Remarks

DPDK Summit San Jose 2016 - August 10th - 11th, 2016

The DPDK community met at The Tech Museum of Innovation in San Jose at a two-day event to discuss the application of DPDK to a variety of industry segments including telecom, cloud, enterprise, security, and financial services. The event enabled the DPDK open source community to share DPDK usage and implementation; to hear from DPDK developers, contributors, and users; and to build the DPDK community.

Jim St. Leger

Software Product Line Manager,
Intel

Introduction

This presentation will outline the roadmap for future DPDK releases including 16.11 and 17.02.

Tim O'Driscoll

Software Engineering Manager for DPDK,
Intel

Roadmap for Future Releases

This presentation will outline the roadmap for future DPDK releases including 16.11 and 17.02.

Hemant Agrawal

NXP

Shreyansh Jain

NXP

DPDK on Embedded Networking SoCs - Experience & Needs

This presentation will focus on NXP experiences in contributing to DPDK, and areas of DPDK that need to be enhanced to improve support for ARM-based processors.

Jerin Jacob

Cavium

Extending DPDK to Add an Event Driven Programming Model

Cavium will provide an overview of event driven programming and the RFC API proposal for extending DPDK to adapt such a model. The presentation will cover introduction to event driven programming model concepts, characteristics of hardware-based event manager devices, RFC API proposal, example use case, and benefits of using the event driven programming model.

Deepak Jain

Network Platform Group,
Intel

High Performance Framework for Symmetric Crypto Packet Processing in DPDK

This presentation will provide an overview of cryptodev framework in DPDK. It will show how both software and hardware crypto accelerators can be used transparently from the application, providing an overview of the framework, its API, a performance analysis and comparisons of software and hardware solutions, and finally an example of NFV use case.

Sowmini Varadhan

Oracle Mainline Linux Kernel Group

User Perspectives on Trying to Use DPDK for Accelerating Networking in End-system Applications

This presentation will describe Oracle's experiences in using DPDK to accelerate I/O for typical end-systems applications. The talk will attempt to generate some discussion about areas where API constructs to provide access to key DPDK features would be valuable to enable an easy transition for typical real-world socket applications.

Sameh Gobriel

Intel

Charlie Tai

Intel

Flow Classification Optimizations in DPDK

This presentation will cover flow table design using DPDK, as well as new algorithmic and hardware optimizations for the  RTE hash library that improves on the lookup and the flow update/insert performance. Furthermore, a new research proof of concept (POC) to optimize the OVS flow lookup using a two-layer lookup technology based on DPDK libraries will be highlighted with some preliminary research results.

Muthurajan Jayakumar
(M Jay)

Intel

Helin Zhang

Intel

Intel® 40G Ethernet Controller Architecture, Application and Performance

This presentation will describe how to achieve maximum I/O performance. It will include an architectural view focusing on the key elements in fast path Rx/Tx, typical application usage scenarios, and methods for optimizing performance.

Eyal Cohen

Silicom

Accelerating SSL and OVS at 100G by Leveraging DPDK

This presentation will describe how to use DPDK together with Intel® Quick Assist Technology and the Intel® Ethernet Multi-host Controller FM10000 Family (FM10K) to achieve 100G throughput and OVS offload.

Dharmraj Jhatakia

GM and Head of DCT,
Happiest Minds

Jessel Mathews

Technical Lead,
Happiest Minds

Virtualization of Network Packet Monitoring Systems Using DPDK

In order to better plan and utilize their networks, Network Administrators need solutions which give them the visibility into the network. Happiest Minds enabled the transformation of a leading Network Packet Monitoring company to co-create the Network Virtual Packet Monitoring system leveraging the key technology innovations like DPDK.

Cristian Dumitrescu

Software Architect,
Intel

NFV Use-case Enablement on DPDK and FD.io

This presentation will describe the development of NFV use cases such as a virtualised provider edge router (vPE) using the DPDK and FD.io projects.

Venky Venkatesan

Intel

Stephen Hemminger

Microsoft

Jerin Jacob

Cavium

Hemant Agrawal

NXP

Sowmini Varadhan

Oracle

Technical Panel

The panel will be comprised of some of the technical experts from the DPDK community. It will involve an interactive Q&A with the audience.

Dave Hunt

Intel

DPDK in a Box

DPDK in a Box is small, low-cost DPDK platform running on a Minnowboard. It's not intended for volume production, but may be useful for universities and independent developers who want to work on DPDK but have a limited budget.

Yuanhan Liu

Intel

DPDK Vhost/Virtio Status

A lot of development effort has been done to DPDK vhost-user/virtio recently, including improving the performance, enhancing the stability and adding more functionality. This presentation will describe some recent enhancements including vhost-user multiple-queue and vhost-user reconnect.

Cunming "Steve" Liang

Intel

Jianfeng Tan

Intel

Scalable High-Performance User Space Networking for Containers

Container-based networking is becoming more and more popular because of the short provisioning time, low overhead, good scalability and reusability. This paper describes virtio for container technology, providing a scalable, high-performance, user space virtual network interface for L2/L3 VNFs.

Dr. Peilong Li

University of Massachusetts, Lowell

Understanding the Performance of DPDK as a Computer Architect

In our experiments, OVS­DPDK can achieve a maximum of 8x throughput increase compared with vanilla OVS. To understand the performance difference, we leverage advanced profiling tools such as Intel VTune Amplifier and Linux perf [4] to investigate in detail what system architecture parameters are affected by OVS­DPDK for achieving the speedups.

Tom Herbert

SDN Group,
Red Hat

DPDK, VPP/FD.io and Accelerated Open vSwitch

In this talk, Mr. Herbert compares VPP with Open vSwitch. Although both VPP and OVS utilize DPDK for data plane acceleration, they are very different in internal architecture and implementation. Mr. Herbert will discuss these differences in the context of various use cases and how performance can vary and how in different uses one may shine while the other may falter.

Sean Choi

Stanford University

PISCES: A Programmable, Protocol-Independent Software Switch

Software switches are typically based on a large body of code, and changing the switch is a formidable undertaking. Instead, it should be possible to specify how packets are processed and forwarded in a high-level domain-specific language (DSL) such as P4, and compiled to run on a software switch. We present PISCES, a software switch derived from Open vSwitch (OVS) DPDK-based implementation, a hard-wired hypervisor switch, whose behavior is customized using P4.

Sangjin Han

UC Berkeley

Christian Maciocco

Intel

Berkeley Extensible Soft Switch (BESS)

This presentation will describe the Berkeley Extensible Soft Switch (BESS).

Mihir Nanavati

University of British Columbia

Decibel: Dense Disaggregated Disks for the Datacenter

In this talk, we take the position that volumes today should represent a core building block of datacenter storage, analogous to virtual machines. In addition to providing a logical block interface, volumes must also provide additional data plane services necessary in multi-tenant environments, such as performance-isolated resource sharing and access control, at the line-speed of expensive non-volatile memories.

Tomoya Hibi

NTT Network Innovation Labs

Yoshihiro Nakajima

NTT Network Innovation Labs

Hirokazu Takahashi

NTT Network Innovation Labs

Change Before You Have to Be Claimed
[from  Change Before You Have To. (Jack Welch)]

In this talk, we share the latest experiment and performance tuning knowledge of scale-out NFV environment in ShowNet of Interop Tokyo 2016. We deployed a set of DPDK-enabled routing VNFs on DPDK-enabled hypervisor vSwitch called Lagopus vSwitch with DPDK vhost-user PMD to examine their performance scalability.

Konstantin Ananyev

Intel

Transport Layer Development Kit (TLDK)

This presentation describes the Transport Layer Development Kit (TLDK), which is an FD.io project that provides a set of libraries to accelerate L4-L7 protocol processing for DPDK-based applications. The initial implementation supports UDP, with work on TCP in progress. The project scope also includes integration of TLDK libraries into Vector Packet Processing (VPP) graph nodes to provide a host stack.

Raja Sivaramakrishnan

Distinguished Engineer,
Juniper Networks

Aniket Daptari

Sr. Product Manager,
Juniper Networks

DPDK in Overlay Networks and How it Affects NFV Performance

One approach to network virtualization is via end-system IP/VPN based overlays. To implement these end-system IP/VPNs, often a kernel based software module is leveraged. However, when the module resides in the host kernel, it incurs a performance penalty. To alleviate these performance penalties, the OpenContrail implementation leveraged DPDK and ported the kernel based distributed forwarding module to the user space.

Hayato Momma

Principal Engineer,
NEC Communication Systems, Ltd.

Innovative NFV Service-Slicing Solution Powered by DPDK

In this presentation, the speaker will talk about:

  • Introduction of 'Service-Slicing-Gateway' that realizes the IoT-service-slicing
  • Why DPDK is necessary
  • How we overcome the issues faced in the past NFV development

Prem Jonnalagadda

Barefoot Networks

Why are Open and Programmable Data Planes Critical to the Future of Networking?

This talk will present the need for programmability and openness of the data plane and the benefits to the networking industry as a whole. Specifically the talk will include details on P4, a high-level, networking domain-specific and open programming language and the ecosystem that is burgeoning around it.

John McNamara

Intel

Getting Your Code Upstream Into DPDK

In you have a simple one-line patch or a full blown Poll Mode Driver this talk will explain how to get that code upstream into DPDK. It will discuss the DPDK community, the mailing list, the patch process, the contributors guides, the ABI policy, code reviews, documentation and other aspects that make up the DPDK ecosystem.

Franck Baudin

Principal Product Manager
OpenStack NF

Anita Tragler

Senior Product Manager
Red Hat Enterprise Linux NFV

Putting DPDK in Production

This presentation will cover:

  • User perspective on DPDK, including consumability and packaging considerations
  • The importance of ABI stability and Long Term Support
  • Security for DPDK guest and host (DPDK vswitch)

Mike Glynn

Program Manager,
Intel

Community Survey Feedback

We conducted a survey of the DPDK community, soliciting input on a variety of topics including DPDK usage, roadmap, performance, patch submission process, documentation and tools. This session will present the results of the survey, which will help to guide the future direction of the project.

DPDK.org and FD.io Community Summit at OPNFV Berlin - June 21, 2016

This event at the Intercontinental Hotel in Berlin brought the DPDK.org and FD.io community together to deliver technical sessions on the application of the DPDK and FD.io projects to NFV, and more specifically to the OPNFV project.

Ed Warnicke

Cisco

Tim O'Driscoll

Intel

John DiGiglio

Intel

Introduction (DPDK, FD.io)

This session will begin with an introduction to the DPDK/FD.io mini summit, including a review of the agenda and logistics. It will also include overviews of the DPDK and FD.io open source projects, showing their linkage to OPNFV.

Thomas F. Herbert

Red Hat

Mark Gray

Intel

Chris Price

Ericsson

NFV Virtual Switching

Data plane and control plane key requirements to address the high performance, reliable, and scalable virtual switching needs within NFV Infrastructure. Review of current and planned functionality within the DPDK and FD.io projects coming together to build NFV ready virtual switching.

Joel Halpern

Ericsson

Frank Brockners

Cisco

SDN Controller/Orchestration/FastDataStacks

The role of Openstack and Opendaylight controller use-cases relevant to the deployment of NFV nodes using virtual network devices and associated topologies. Review of the OPNFV FastDataStacks project proposal, timeline, and components (Openstack, Opendaylight, and FD.io Vector Packet Processing integration).

Keith Wiles

Intel

Lingli Deng

China Mobile

Data Plane Acceleration

The goal of the Data Plane Acceleration (DPACC) project is to specify a general framework for VNF data plane acceleration, including a common suite of abstract APIs at various OPNFV interfaces, to enable VNF portability and resource management across various SOCs and standard high volume server platforms. This session will review the requirements being defined by the DPACC project, and will describe the enhancements that have been made to the DPDK framework to satisfy those requirements.

Carlos Goncalves

NEC

Maryam Tahhan

Intel

Service Assurance

A key requirement for NFV solutions is to be able to provide Service Assurance, including the ability to measure and enforce traffic quality KPIs, as well as detecting and reporting any violations to higher level EMS/OSS systems. Within OPNFV, these needs are being addressed by the Software Fastpath Service Quality Metrics (SFSQM) project. This session will describe the requirements that have been identified within the SFSQM project, and the enhancements that have been made to the DPDK framework to support Service Assurance.

Chris Donley

Huawei

Roadmap/Feedback Discussion

Attendee participation expanding on the subject areas reviewed during the day as well as introducing other possible OPNFV project intersection points.

DPDK Summit China/Asia Pacific 2016 - May 18, 2016

The DPDK community met at the Renaissance® Shanghai Yangtze Hotel to discuss the application of DPDK to a variety of industry segments including telecom, cloud, enterprise, security, and financial services. The event enabled the DPDK open source community to share DPDK usage and implementation; to hear from DPDK developers, contributors, and users; and to build the DPDK community.

Jim St. Leger

Software Product Line Manager,
Intel
Jim St. Leger is a software product line manager in Intel’s Network Platforms Group. His work encompasses software that accelerates network traffic and applications, focusing on improving the speed and efficiency of packet-based networks through the use of multicore processors and software technologies. Jim is part of the team that started the Data Plane Development Kit (DPDK.)  He is also involved in Hyperscan, a regex matching engine (01.org/hyperscan), and the Fast Data (FD.io) packet processing open source projects.

DPDK Community Update and an Introduction to the Fast Data, FD.io, Project

The DPDK Community and open source software project has been growing steadily for the past four years. This session will look at the growth of the community in terms of contributors and commitments, discuss who is involved and contributing to the community today, and provide guidance on how everyone and anyone can start contributing to DPDK.org today. The Fast Data or FD.io Project launched in February 2016.  This session will provide the background on the open source project creation including the gap it fills in the NFV/SDN data plane, packet processing capability stack.  It will also talk about the rapid initial growth of the project including membership and future direction.

Huawei Xie / 谢华伟

Software Engineer,
Intel
Huawei Xie joined Intel in 2006 and now is leading the development and performance optimization of the DPDK virtualization data plane. He has worked in various areas of NFV, full functional platform simulator, performance analysis and projection. He received a MS degree in Computer Science from Xi’an JiaoTong University in 2006, and a BS degree in Chemical Engineering from Tsinghua University in 2003.


Jianfeng Tan / 谈鉴锋

Software Engineer,
Intel
Jianfeng Tan joined Intel in 2015. Before Intel, he worked for VMware on the NSX product for one year. Jianfeng is interested in using DPDK to accelerate container networking. Jianfeng received a MS degree in Computer Science and Technology from Tsinghua University in 2014, and a BS degree in Software Engineering from Nankai University in 2011.

Accelerate virtio/vhost Using DPDK in NFV/Cloud Environment

As the standard para-virtualization interface, the performance and stability of virtio and vhost are the key to the success of NFV. In this presentation, we would like to summarize our many years of pioneering work around DPDK virtio/vhost. We introduce the performance optimization techniques around virtio ring layout and vhost TSO to accelerate TCP/IP stack based applications in the guest VM. To enhance the robustness of vhost, we created ‘vhost reconnect’ to support vhost restarting and reconnecting to QEMU in the event of a crash. We also leverage VMFUNC to provide protected and fast inter-VM channels. To provide high throughput and low latency interface in the container environment, we address the device simulation and translation gap and successfully created virtio interface in the container.

Jun Xiao / 肖骏

Founder & CTO,
CloudNetEngine
Jun Xiao is the founder and CTO of CloudNetEngine which is focused on innovating the next generation "engine" for cloud virtual networking. He has more than 15 years experience in IT/Telco companies such as VMware, Huawei, Lucent, and Sun Microsystems. Jun has extensive system design experience ranging from low level device drivers, network stacks, to middleware software and distributed systems. Jun Xiao holds more than 10 US/China issued/pending patents and is one of key inventors of VXLAN.

Next Gen Virtual Switch

With the increasing prevalence of cloud computing, there is a proliferation of east-west traffic in clouds, and growing demands for virtualizing network I/O intensive workloads at telcos with big data. All those put huge amount of challenges on existing virtual switches. In this presentation, we will share with the DPDK community what we learned while building the CloudNetEngine virtual switch, which is based on great open source projects like DPDK, OVS, NPF, etc. We'll have deep dive discussions on how to boost performance, improve CPU efficiency, implement a rich feature set needed by public and private clouds — overlay, security group, QoS, HW/SW offload, load balance, monitoring and ease of integration with cloud ecosystem.

Dr. KyoungSoo Park

Associate Professor Electrical Engineering,
KAIST
KyoungSoo Park is an associate professor in the Electrical Engineering department at KAIST. He received his B.S. degree from Seoul National University in 1997, and his M.A. and Ph.D. degrees from Princeton University in 2004 and 2007, respectively, all in computer science. Before joining KAIST, he was an assistant professor in the computer science department at University of Pittsburgh in 2009. His research is focused on the reliability, performance, and security issues in the design and implementation of networked computing systems. He developed CoBlitz, a scalable large-file content distribution network (CDN), which was acquired by Verivue, Inc., and subsequently by Akamai, Inc. in 2012. He co-developed HashCache, a memory-efficient caching storage system for developing regions, which was chosen as one of the top 10 technologies in 2009 by the MIT Technology Review Magazine. Most recently, he co-authored a mTCP paper, which received the Community Award at USENIX NSDI 2014. His recent research is focused on high performance packet/flow processing systems using multicore/manycore processors, such as PacketShader (a 40 Gbps software router), SSLShader (13 Gbps SSL proxy), and Kargus (33 Gbps software IDS).

mTCP: A High-Speed User-Level TCP Stack on DPDK

Scaling the performance of short TCP connections on multicore systems is fundamentally challenging. Despite many proposals that have attempted to address various shortcomings, inefficiency of the kernel implementation still persists. For example, even state-of-the-art designs spend 70% to 80% of CPU cycles handling TCP connections in the kernel, leaving very little room for innovation in the user-level program. In this talk, I will present mTCP, a high-performance user-level TCP stack for multicore systems. mTCP addresses the inefficiencies from the ground up — from packet I/O and TCP connection management to the application interface. In addition to adopting well-known techniques, our design (1) translates multiple expensive system calls into a single shared memory reference, (2) allows efficient flow-level event aggregation, and (3) performs batched packet I/O for high I/O efficiency. Our evaluations on an 8-core machine showed that mTCP improves the performance of small message transactions by a factor of 25 compared to the latest Linux TCP stack and a factor of 3 compared to the MegaPipe system. It also improves the performance of various popular applications by 33% to 320% compared to those on the Linux stack.

Lou Yang / 娄扬

Software Architect,
TOPSEC
Lou Yang graduated from Dalian University of Technology in 2001. He joined TOPSEC in 2005 and now is working on the next generation firewall. Lou Yang has 15 years experience in network software system design and development. He has participated in or guided several large scale network system software projects.

DPDK: A Journey of Migration to Linux Kernel

In some special cases, we can not use DPDK directly, for instance, if our software runs in linux kernel mode or we use a special software architecture. But we can migrate DPDK’s key elements into our own products. The key elements include efficient memory management, polling mode NIC driver, undisturbed data plane, etc. This talk will introduce the practices of how to migrate DPDK into a network firewall product.

Dr. Kai Zhang / 张凯

Ph.D.,
University of Science and Technology of China
Dr. Kai Zhang received his Ph.D. in Computer Science and Technology from the University of Science and Technology of China in 2016. He was a Visiting Scholar at Ohio State University during 2013 and 2015. His research interests include networked systems, high performance computing, and big data.

Building High-Performance Networked Systems with Innovative Software and Hardware

As network I/O speed has been unleashed by new software techniques such as DPDK, network processing is no longer the bottleneck of networked systems. Consequently, networked systems need redesign to meet the increasing demand for high speed network processing. In this talk, we make a strong case for GPUs to serve as special-purpose devices to greatly accelerate the operations of in-memory key-value stores. Specifically, we present the design and implementation of Mega-KV, a GPU-based in-memory key-value store system that achieves high performance and high throughput. Effectively utilizing the high memory bandwidth and latency hiding capability of GPUs, Mega-KV provides fast data accesses and significantly boosts overall performance. Running on a commodity PC installed with two CPUs and two GPUs, Mega-KV can process up to 160+ million key-value operations per second, which is 1.4 - 2.8 times as fast as the state-of-the-art key-value store system on a conventional CPU-based platform.

Xu Liang / 徐亮

Director,
UCloud.cn
Xu Liang has 15 years of software development experience in the telecommunications, Internet and data communication fields. He joined Cloud in 2015 and now is working on the large scale virtual network and the next generation of high-performance NFV application.

Vortex from UCloud

As an IaaS company we developed many NFV applications via DPDK, but just recently we released Vortex, a Layer-4 load balancer. We will discuss the challenges in developing a large-scale, multi-tenant and Overlay Network NFV application and how we handled these challenges with DPDK. We will also discuss our experience and lessons learned.

Wang Zhihong / 王志宏

Software Engineer,
Intel
Wang Zhihong is a software engineer of Intel, currently working on DPDK virtualization performance analysis and optimization. He is a graduate of East China Normal University, where he received his master's degree in computer science in 2011.

A Deep Dive Into Memory Access

Memory efficiency is critical to VNF performance. It is challenging to design and implement memory friendly networking software, especially in a multiple core/processor environment. Understanding of the microarchitecture and underlying memory hierarchy helps software architects and developers analyze and optimize software performance.
This session uses a DPDK based NFV example to illustrate the actual memory behavior behind software abstraction & C code and common optimizing techniques. It also uses CPU cycles to explain the overhead of each type of memory operation to give a sense of what should be avoided and what's the right thing to do in practice.

Dr. Dan Li

Associate Professor in Computer Science,
Tsinghua University
Dan Li is an Associate Professor in the Computer Science Department of Tsinghua University. He received his M.E. and Ph.D from Tsinghua University in 2005 and 2007 respectively, both in computer science. His current research include Future Internet architecture, software defined networking and cloud computing systems. He has published more than 50 technical papers in referred conferences and journals, including ACM SIGCOMM, IEEE ICNP, IEEE INFOCOM, ToN, TC, TPDS, etc. He served as an associate editor of IEEE Transactions on Computers (2013-2015).

Light and NOS

We designed and implemented a user-level network stack based on DPDK, named Light. The benefit of Light is that it does not need the application to modify anything, and the protocol stack does not affect the performance of the application.

Haomai Wang / 王豪迈

CTO,
XSKY
Haomai is an active Ceph developer and contributor to several open source projects. He specializes in storage, filesystem, database and virtualization. He was previously an OpenStack contributor and maintains several OpenStack cloud systems.

When Ceph Meets DPDK

In this presentation we will discuss the integration with DPDK, SPDK and Ceph. Ceph is a popular open-source storage system which includes block, file, and object interfaces. We implement a new DPDK network stack in Ceph which contains userspace TCP/IP stack. SPDK is another Intel open-source technology which implements Userspace NVME protocol. With MBUF from DPDK, we make the whole data packet without copy to store to NVME device. The whole Userspace stack is numa friendly, zero copy and nearly lock free.

DPDK Userspace 2015 Dublin, Ireland - October 8 - 9, 2015

Userspace 2015 was a developer forum which focused on the elements of DPDK which are most pertinent to the open source software community members. The two-day event at the Ballsbridge Hotel in Dublin included highly interactive discussions on the latest features and upcoming changes to DPDK.

Thomas Monjalon

Packet Processing Engineer and DPDK.org Maintainer,
6WIND
Thomas Monjalon is a packet processing engineer driven by Open Source and working in a great company which thinks that speed matters. When DPDK was released as a zip file, he started sharing code and tools on a dedicated website, which helped to gather and strengthen the community which is still growing.

Project Growth and Next Steps

Thomas presented on the current role of being a maintainer of DPDK in the community, the rapid speed of growth in the project and the future of the dpdk.org community.

Cristian Dumitrescu

Software Architect,
Intel
Cristian Dumitrescu is a SW architect within Network Products Group (NPG) at Intel Corporation. His current focus is efficient implementation of networking workloads on Intel CPUs and the Data Plane Development Kit. He holds a BSc, MSc and PhD from Politehnica University of Bucharest, Romania.

DPDK Packet Framework

Cristian presented the latest evolution of the DPDK Packet Framework, how it can be used and future proposals for the extension of the functionality of Packet Framework.

Thomas Herbert

Principal Software Engineer,
Red Hat
Thomas F Herbert is a Principal Software Engineer with Red Hat. He has a background ranging from embedded to networking. He is Author of "Linux TCP/IP Networking for Embedded Systems." Tom has developed SDN applications for Service Provider networks and network security. Currently at Red Hat in the SDN team, he contributes to to Open NFV (OPNFV-OVS,) Open vSwitch (OVS) and DPDK.


Mark Gray

Software Engineer,
Intel
Mark Gray is a Software Engineer at Intel Corporation based in Shannon in the West of Ireland. Mark works on software dataplane technologies with a focus on virtual switching. Mark was the maintainer of Intel's DPDK-Accelerated fork of the Open vSwitch project commonly called OVDK and is currently working on the userspace datapath in the Open vSwitch project.


Kevin Traynor

Software Engineer,
Intel
Kevin Traynor joined Intel in 2004, having previous worked on Bluetooth and IC development. During his time at Intel he has worked on a variety of voice, security and packet processing projects. In his current position he is involved in the design and development of features and performance improvements for Open vSwitch with DPDK.

OVS, DPDK and Software Dataplane Acceleration

Virtual Switching with DPDK was discussed by trio Thomas, Mark and Kevin from two different viewpoints in this presentation on the current challenges and opportunities of integrating DPDK with Open vSwitch technologies.

Venky Venkatesan

DPDK Architect,
Intel
Venky Venkatesan is a Senior Principal Engineer in the Network Platforms Group at Intel, where he guides the overall effort to make networking infrastructure functions run efficiently on Intel Platforms. His current area of focus is the Data Plane Development Kit, which is a project he has guided within Intel since its inception. Venky has been with Intel for 20 years and holds a Bachelor’s in Computer Engineering from the University of Mumbai, India.


Bruce Richardson

Software Engineering
Technical Lead,
Intel
Originally from Sligo in the north-west of Ireland, Bruce Richardson, enjoys making absolutely terrible puns and talking about himself in the third person. He's also passionate about his food - a statement that some may find hard to swallow [pun absolutely intended] - and has been working on DPDK for over 2.3 x 10^15 packet times (64B @10GB Line rate, obviously!). When not helping to save the world, one unprocessed packet at a time, he's kept busy by his three little minions/boys at home.

The 7 Deadly Sins of Packet Processing

Venky and Bruce paired up for this presentation on the pitfalls often faced in Packet Processing and the optimum use of DPDK.

Andras Kovacs

Lead Software Developer,
Ericsson
András graduated in 1998 in the field of Electrical Engineering. Currently working on Ericsson's high performance platform which provides the packet forwarding infrastructure for the Ericsson Virtual Router (EVR).

During the 15 years he spent with Ericsson worldwide (Hungary, Brasil, Sweden, Canada, Australia, USA), he was involved in the development cycles of several products and different technologies including Fixed-line Networks (Public Switching Telephony Networks), Ericsson's ATM switch (AXD), and IP Routers. He has been working in the packet processing area for 6 years now, going through different architectures, home grown packet processing ASICs, Network Processors and now packet processing on x86 (2.5 years).


László Vadkerti

Lead Software Developer,
Ericsson
László is currently working on Ericsson's high performance platform which provides the packet forwarding infrastructure for the Ericsson Virtual Router (EVR).

During the 15 years he spent with Ericsson worldwide (Hungary, Brazil, Sweden, India, USA), he was involved in the development cycles of several products and different technologies including Fixed-line Networks (Public Switching Telephony Networks), Ericsson's ATM switch (AXD), Mobile Media Gateway and IP Routers. He has been working in the packet processing area for 11 years now, going through different architectures, like special purpose Network Processors, home grown packet processing ASICs and now packet processing on x86 (3 years).

Generic Resource Manager

László & András (Ericsson) presented thoughts on proposed Generic Resource Manager and Memory management in DPDK.

Andy Harvey

Distinguished Engineer,
Cisco Systems
Andy is a Cisco Distinguished Engineer. Since joining Cisco in 1993, Andy has been instrumental in such key products as the 1000, 1600, 2500, and 5300 and the Integrated Service Routers (ISR) family. He had a brief stopover in Network Management whereas the Architect of Cisco's Network Services (CNS) he promoted a visionary approach for model driven programmatic networks. He is currently working with the CSR100v Virtual Router team in the area of IO, Virtualization and Performance.

DPDK Architecture Musings

Andy brought forward the user perspective of integrating DPDK with applications, how the community can help users to develop applications faster and more easily in real world use cases.

Roger Melton

Technical Leader,
Cisco Systems
Roger joined Cisco Systems in 1999 and is a Senior Technical Leader in the Enterprise Routing Group working on data plane technologies. He has 30 years’ experience as an embedded software developer and architect. His current focus is data path I/O for the Cisco Cloud Services Router (CSR-1000v), a virtual router that runs Cisco IOS-XE software on various hypervisors including KVM, Hyper-V, Xen and Vmware.

DPDK Integration Journey

Roger presented his experience of DPDK from an Architecture perspective reflecting on the opportunities and challenges facing the community from a technological perspective as the project continues to scale.

Hanoch Haim

Principal Engineer,
Cisco Systems
Hanoch is a 20 year Telecom Industry veteran working for Cisco Systems. Most recently, as a Principal Engineer in the Services Group. Hanoch has vast experience and deep knowledge in embedded systems, advanced HW/SW technologies and complex SW systems. Hanoch has been key player in NBAR2/Deep packet inspection project and was responsible for many of the innovative initiatives. Hanoch has initiated, developed and is leading a virtual team developing the TRex a stateful traffic generation tool that is enhancing Cisco ability to test state-full features.

T-Rex Stateful Packet Generator

Hanoch Haim presented on TRex Stateful Packet Generator, an open source, low cost, stateful traffic generator fuelled by DPDK. It generates L4-7 traffic based on pre-processing and a smart replay of real traffic templates. TRex amplifies both client and server side traffic and can scale to 200Gb/sec with one UCS.

Bill Fischofer

Technical Lead,
Open Data Plane
Bill has over 25 years’ experience in the information technology industry working on everything from IBM mainframes to ASICs. Prior to joining the Linaro Networking Group in September 2013 he spent the previous decade at several private and public companies designing and programming data plane firmware for several generations of networking SoCs. Bill is currently the LNG technical lead for the OpenDataPlane project.

OpenDataPlane: A Quick Introduction and Overview

Bill presented on the Linaro Networking Group’s OpenDataPlane which supports application portability across diverse ISA and system architectures.

Mohammad Abdul Awal

Software Engineer,
Intel
Mohammad Abdul Awal is working as a software engineer in Intel Shannon, begginning his professional career in 2001. Before joining Intel, he worked for Ericsson and mainly focused on probe-based QoS/QoE monitoring for VoIP applications using DPI. He also worked in probe-based TCP performance monitoring, Voice over LTE monitoring. He also worked in mobile ad-hoc networks for distance learning and disaster recovery. He obtained his PhD degree in wireless communication from University of Paris-Sud, France in 2011. He also obtained his masters and bachelor degree in computer science from Asian Institute of Technology, Thailand in 2005 and American International University Bangladesh in 2001, respectively. He has written more than 15 international journal and conference papers and one patent.

Hyperscan Software Pattern Matching

Awal describes Hyperscan a software-based regular expression matching library, supporting large-scale, high-performance, streaming regular expression matching on Intel Architecture.

Declan Doherty

Software Engineer,
Intel
Declan Doherty is a software engineer in Intel Shannon, focusing on the Data Plane Development Kit. Prior to joining Intel, Declan worked as a member of performance/infrastructure team in an ultra-low latency financial trading group, and as part of a development team building a range of real time applications and device drivers for a carrier grade optical distributed layer 2 switch. He has a degree in Electronic Engineering for the National University of Ireland Maynooth.

Within the DPDK project, Declan is the author of the new cryptodev API and is maintainer for the link bonding library.

A Symmetric Cryptography Framework for DPDK

An introduction to the new asynchronous burst oriented symmetric Cryptography API and device framework for DPDK.

Stephen Hemminger

Principal Software Architect,
Brocade
Stephen Hemminger has been actively involved with Linux kernel development for over 10 years. He has contributed several network device drivers, a network emulator, and is the maintainer of the bridging and iproute2 utilities. After leaving the Linux Foundation, he went on to Vyatta (now acquired by Brocade) where he helped develop their open source router operating system. In his current position at Brocade, he is actively involved in design and development of the high performance vRouter using the Dataplane Development Kit (DPDK).

DPDK Performance Lessons Learned in vRouter

Stephen presented on DPDK performance lessons learned in developing Brocade’s vRouter, including their experiences with QoS and LPM.

Tetsuya Mukawa

Software Engineer,
IGEL Co. Ltd.
Tetsuya Mukawa is a software engineer at IGEL Co. Ltd. and one of the developers of Lagopus, an open source software switch. He has roughly 10 years’ experience in embedded Linux products. His interests include high performance software switches and VNF.

Dynamic NFV Deployment with Port Hotplug and Virtio

Tetsuya presented on port hotplug and how it can be used to support dynamic NFV deployment.

DPDK Summit San Francisco 2015 - August 17, 2015

The DPDK community met at the Westin St. Francis Hotel in San Francisco to discuss the application of DPDK to a variety of industry segments including telecom, cloud, enterprise, security, and financial services. The event enabled the DPDK open source community to share DPDK usage and implementation; to hear from DPDK developers, contributors, and users; and to build the DPDK community.

Tim O'Driscoll

Software Engineering Manager for DPDK,
Intel
Tim is a software engineering and program manager overseeing the DPDK project and development activities at Intel. He has a strong telecom background having spent his career working at Ericsson and Motorola in positions including engineering and system and solution architecture before coming to Intel. He’s originally from Ireland and has traveled extensively including multiple posts around the globe.

Opening Remarks and Kickoff to DPDK Summit

On August 17, 2015, Tim O’Driscoll, an Engineering Manager from Intel, provided the opening remarks to kick off the DPDK Summit 2015.

Tim Mortsolf

CTO and Co-Founder,
RIFT.io
Tim Mortsolf is a co-founder and the chief technology officer of RIFT.io. As CTO, Mortsolf helps define the company’s technology strategy and works directly with the engineering team to develop innovative product solutions. Mortsolf worked in software research and development fields of internet routing and subscriber management throughout his career at Bell Laboratories, U.S. Robotics, RedBack Networks and Starent Networks. Mortsolf co-founded and served as the chief architect at Affirmed Networks. Mortsolf holds a large number of issued patents in the telecommunications and internet technology fields and is an author of several Internet Engineering Task Force (IETF) networking standards. At U.S. Robotics, Mortsolf defined the Point-to-Point Tunneling Protocol (PPTP) that enables secure remote access to corporate networks across the public Internet. For his work in this field, he was a recipient of the PC Magazine Networking Software Innovation of the Year award. Mortsolf is a summa cum laude graduate from the University of Massachusetts and holds several degrees in physics, chemistry, as well as biochemistry and molecular biology.


Scott Myelle

VP Solutions Architecture,
RIFT.io
Scott Myelle is a 20 year communication industry veteran with experience creating solutions and implementing them. At RIFT.io he has responsibility for Solution Architecture, working closely with customers to evangelize RIFT.io’s vision and understand their future needs and requirements. Throughout his career Scott’s passion has been on evolving networking architectures from the days of TDM through the most modern cloud designs. Prior to RIFT.io Scott managed a group of Consulting Solution Architects at Cisco System focused on the intersection of cloud, virtualization and software defined networking. Prior to Cisco, Scott held several management positions at Contextream (acquired by HP), Tellabs and United Pan-Europe Communications (acquired by Liberty Global).

Leveraging DPDK to Scale-Out Network Functions Without Sacrificing Networking Performance

NFV application workloads are deployed in ecosystems with varying network attachment conditions that determine the availability of specific DPDK technologies. DPDK technology has rapidly evolved to support multiple I/O models ranging from dedicated access with PCI pass-through, shared access with SRIOV, and vhost-user offload using a DPDK enabled Open vSwitch. This presentation demonstrates how to write a flexible network function that can utilize DPDK to its full potential while retaining the ability to run in a non-DPDK environment.

Charles Shiflett

Senior Software Engineer,
IBM Aspera Solutions
Charles Shiflett is a senior software engineer at Aspera, an IBM company. Shiflett works as the lead developer and architect for Aspera’s Next Generation FASP. Next Generation FASP is a high speed (80Gbps), adaptive, secure and reliable congestion controlled transfer protocol built upon Aspera’s FASP transfer protocol. Shiflett has a Bachelor of Science in computer engineering from UC Santa Cruz, and a slight obsession with improving transfer I/O.

Aspera’s FASP Protocol Uses Standard Hardware and DPDK to Achieve 80Gbps Data Transfer

Charles Shiflett will review the technologies and design approach to send data at a rate in excess of 1 TB every two minutes. Aspera Fast, Adaptive, and Secure Protocol (FASP®) is a breakthrough transfer protocol that leverages existing WAN infrastructure and commodity hardware. Code samples showing the use of DPDK, AES-NI, FASP Sockets, and direct I/O to create a zero-copy transfer technology will be discussed.

Keith Wiles

Staff Architect,
Intel
Keith Wiles is a staff architect at Intel Corporation working with DPDK and NFV acceleration technologies for accelerated networking performance. Wiles serves as a contributing member in the Data Plane Acceleration group (DPACC) within OPNFV. Wiles also authored Pktgen-DPDK, a network traffic generator running on DPDK. He wrote Pktgen-DPDK while working at Wind River to understand DPDK and introduce DPDK as part of the network acceleration platform products within Wind River. He worked at Wind River for 16 years while enhancing VxWorks Real Time networking stack. Wiles designed and wrote the CertStack a DO-178B certified TCP/IPv4 network stack used in the MILS and AIRINC platforms. Wiles was also a founding member, chief technology officer and chief financial officer of XAct Inc before Wind River acquired the company that produced software and hardware for Ethernet managed and unmanaged switch designs.

Future Enhancements to the DPDK Framework

This session will provide insight and gather community input on forward-looking activity in advancing DPDK to include connectivity to hardware accelerators and SOC support. Keith will review the need for additional devices and functionality within the DPDK framework including supporting non-PCI configuration, external memory manager and event programming model utilized by many SOCs. The session will drill down on the use of a crypto device within the DPDK framework by reviewing an early proof of concept of a software and hardware implementation of the device.

Yoshihiro Nakajima

Researcher,
NTT Network Innovation Laboratories
Yoshihiro Nakajima received his Bachelor of Science in information science in 2003, a Master of Engineering degree in 2005, and the doctorate degree in computer science in 2008 from the University of Tsukuba. He worked as a research fellow at the Japan Society for the Promotion of Science from 2005 to 2008. He joined NTT Network Innovation Laboratories in 2008 and studied high-performance stream processing systems and a multi-layer network management system. He is a project lead for a high-performance SDN/OpenFlow switch called Lagopus vSwitch. His interests include high-performance software packet processing middleware and technologies. He is a member of the Association for Computing Machinery (ACM) and Institute of Electrical and Electronics Engineers (IEEE).

It's Kind of Fun to Do the Impossible with DPDK

NTT Network Innovation Laboratories will present lessons learned from a one year experiment on SDN/OpenFlow Lagopus Switch development and trials on ShowNet SDN-IX from Interop Tokyo 2015. A co-design of FPGA NIC, DPDK library extension and software data plane is indispensable to improve packet lookup/processing performance and to reduce CPU resources for 100Gbps packet processing performance. Additionly, NTT will share a carrier use case activity on hybrid SDN with autonomous network control and network policy control by their Lagopus switch and OpenFlow technologies.

Arun Rajagopal

Technology Architect,
NFV and Wireless Core,
Sprint
Arun Rajagopal works in the chief technology officer’s office at Sprint and manages technology strategy and architecture for NFV. In his current role, he defines the long term architecture for NFV, develops architecture evolution plans and creates requirements for NFV adoption at Sprint. This includes looking at global mobile technology requirements and trends and creating strategies for adoption or adaption to meet Sprint’s needs over the medium to long term. Prior to working at Sprint, Rajogopal was the chief network architect at Reliance Globalcom and responsible for strategy, architecture and design for Reliance’s global IP network that spans over 200 countries. Rajagopal has a Bachelor of Technology in electronics and communications engineering from Calicut University in India and a Master of Computer Engineering from Texas A&M University.


Sameh Gobriel

Senior Research Scientist,
Intel Labs
Sameh Gobriel is a senior research scientist at Intel Labs where he drives research to enable future products to be best-in-class in energyefficient performance. His research interests include platform I/O architecture, software networking, Network Functions Virtualization and energy-efficient design of computer platforms.

He received his BE in Electronics and Electrical Communications Engineering from Cairo University in 1999, the MS and the PhD degree in computer science from the University of Pittsburgh in 2007 and 2008, respectively.

Dr. Gobriel is the author of more than 30 research papers, articles and book chapter in first tier conferences and journals; he has 28 filed technology patents. For his outstanding research work at Intel Labs he has received a few career awards including an Intel Achievement Award for his research on Energy-Efficient WiFi Interfaces. He has served on the program committees of numerous conferences and workshops.

Design Considerations for a High-Performing Virtualized LTE Core Infrastructure

Sprint’s expectation is to achieve similar performance in moving from purpose built ASIC based platforms to virtualized network solutions running on high volume servers. This session will discuss the technical challenges in achieving a scalable solution that addresses the required transaction rates and throughput of a carrier network. Learn how DPDK, VM to VM communication optimizations, and cluster scaling technologies work together to create a scalable LTE core infrastructure built on high volume servers.

Al Sanders

Lead Developer,
Hewlett-Packard
Al Sanders works in part with the Hewlett-Packard (HP) Server Telco team managing the NFV Infrastructure Lab. The lab partners with NFV technology providers to ensure that HP Server platforms provide optimal NFV Infrastructure and packet processing performance. Sanders worked at HP for 33 years and worked on Linux solutions for Telco customers for the past 8 years. Previous assignments include work on HP-UX kernel drivers, DCE RPC and DFS environments as well as hardware simulation tools. Sanders graduated from the Massachusetts Institute of Technology in 1982 with a Bachelor and Master of Computer Science.

Evaluation and Characterization of NFV Infrastructure Solutions on Hewlett-Packard Server Platforms

The HP Servers NFV Infrastructure Lab was formed to evaluate DPDK based environments to ensure that HP Server Platforms can provide the best possible performance for hosting NFV Solutions. The lab has partnered with a number of NFV providers, including Intel’s Open Network Platform and 6WIND. Our testing methodology will be presented with a focus on packet processing throughput and latency in a variety of DPDK enabled configurations, including bare metal, SR-IOV, and accelerated virtual switches. Examples of results using Intel’s ONP and 6WIND technologies will be presented.

Jim St. Leger

Software Product Line Manager,
Intel Network Platforms Group
Jim St. Leger is a software product line manager in Intel’s Network Platforms Group. His work includes software that accelerates network traffic and applications as well as improving the speed and efficiency of packet-based networks through the use of multicore processors and software technologies.

Prior to Intel, St. Leger worked in various manufacturing, product engineering and marketing roles in the automotive, aerospace and semiconductor equipment industries. He holds a Bachelor of Science in mechanical engineering from Rensselaer Polytechnic Institute (a.k.a. RPI) and a Master of Management in manufacturing (MMM) from Northwestern University (McCormick School of Engineering & Kellogg School of Management.)

St. Leger spends his off-hours focused on his family, the outdoors, lifelong learning and a myriad of other random pursuits. He works as an advisory board member to Hacking Autism, mentors a local FIRST Robotics team and chairs the KAET Arizona PBS Community Advisory Board. Follow him on Twitter: @Intel_Jim.

Open Discussion Panel (Q&A with Speakers)

Jim St. Leger, Intel’s Software Product Line Manager, led a discussion with DPDK experts. During this time, DPDK Summit attendees had the opportunity to ask detailed questions of the day’s presenters.

DPDK Summit China 2015 - April 21, 2015

The DPDK community met at the JW Marriott Hotel in Beijing to discuss the application of DPDK to a variety of industry segments including telecom, cloud, enterprise, security, and financial services. The event enabled the DPDK open source community to share DPDK usage and implementation; to hear from DPDK developers, contributors, and users; and to build the DPDK community.

Lynn Comp

Director of Market Development Organization,
Network Platforms Group,
Intel
Lynn Comp brings a diversity of experience across multiple digital services touch points, spanning software, datacenter network and communications infrastructure, cellular telephony, and multiple technologies that increase operational automation and efficiency in cloud datacenters. Currently Lynn’s organization is responsible for customer, communications service provider, and ecosystem partner collaboration to accelerate and simplify the transformation of the network.
Lynn has a BSEE from Virginia Tech and an MBA from University of Phoenix.

Kick - off Speech

Lynn starts off the summit by discussing the objectives, DPDK history, the community, its contributors, and real world DPDK deployments.

Dr. Deng Hui

China Mobile Dr. Deng Hui obtained his PhD degree from the Dept. Computer Science and Technology at Tsinghua University in 2002. He joined China Mobile in 2007 and now is working on the evolvement of the operator's architecture during the Mobile Internet Era. Hui Deng has been acting as the co-chair of IETF MIF working group since 2008. Additionally, he used to be a board member of Wireless Broadband Alliance (WBA) and the chair of Program Advisory Group within the board between 2012 and 2014. He has been a member of the board of directors of OPNFV since 2014.

Accelerate NFV with DPDK

Network Function Virtualization (NFV) aims to enable network operators to architect networks by evolving standard IT virtualization technology to consolidate network equipments onto industry standard high volume servers. The performance requires that the data plane devices within network infrastructure have been growing at significantly higher rates.

Johnson Liu

Principle Forwarding Architect,
Huawei Technologies Co., Ltd.
Johnson has sixteen years experience in IP R&D and was mainly engaged in network software system design and development. As a chief designer, he has guided several large scale network software platforms. Johnson is skilled at the following fields: IP protocols and application, router system and software, high performance forwarding system and NFV.

Transforming the Data Plane in the Telco Environment

In Johnson's speech, he will introduce the key requirements and philosophies of data plane software system, give an analysis of the DPDK architect, and then introduce DPDK application and issues in its development. He will also give some viewpoints in PMD decoupling with DPDK framework, PMD API and community operations of DPDK.

Liang Cunming

DPDK Tech Lead,
Intel China Ltd.
Cunming has 10+ years packet processing experience. He has been a DPDK developer since 2011 and now is the DPDK tech lead in Intel, China. His recent contributions to DPDK.org include the multi-thread per core, bifuricated driver and Vector PMD. He is an expert in DPDK PMD, IO virtualization and SIMD optimization. He is passionate about making DPDK for customers of the PRC and the rest of the world.

Best Practices for Building Core/Efficient Applications

This provides a new perspective to the DPDK core efficiency problems. For a specific packet processing workload, the core efficiency means two sides: the more effective cycles and the less idle cycles. This talk will introduce the practices of using SIMD instruction to reduce the effective CPU cycle cost of packet IO. The two different methods of handling the idle cycles by using the power management or by preemptive scheduling will also be discussed.

Dr. Ou Liang

Program Manager of Data Network Communication Division,
China Telecom Guangzhou Research Institute
Dr. Ou is a program manager of Data Network Communication Division, China Telecom Guangzhou Research Institute (GSTA). Currently, he is working on enabling SDN technologies in fixed broadband IP network. Prior to this, he was responsible for R&D work on intelligent carrier grade metro network, planning system for IP backbone network and innovative networking services based on Internet technologies. He obtained a Ph.D. degree in electronics and information engineering from Huazhong University of Science and Technology, Wuhan, China in 2005.

SDN/NFV Based IP Edge Solution on Intel ONP

1. States the transformation requirements of IP network for legacy telecom carriers on the background of SDN.
2. Gives a brief explanation on CT's Programmable Forwarding Architecture (PFA) and NFV solution based on DPDK.
3. Introduce recent cooperation R&D work between CT and Intel on the basis of Intel ONP.
4. DPDK use cases and testing results.

Dr. Zhu Chao

Researcher,
IBM-research China
Dr. Zhu Chao graduated from Institute of Computing Technology, Chinese Academy of Sciences. He has years of experience on wireless network cloud. His team built the world's first multi-mode NFV for telecom prototype which was shown on MWC2014. He is interested in many areas, such as I/O virtualization, network packet processing, cloud computing and so on.

DPDK on Power: A New Enhanced Packet Processing Architecture

In this talk, a new computer architecture which can support DPDK - IBM Power will be introduced. The speaker will share his experience on the DPDK multiarchitecture migration. Also, he'll show some DPDK use cases on Power.

Sun Chenghao

Senior Software Engineer,
Alibaba
Chenghao currently is a technical expert for the Alibaba data center networking infrastructure team. He has rich R&D experience in Firewall, Switch and Software Gateway. At present, he is concentrating on the high performance networking node and virtualization application.

Optimize Cloud Infrastructure with DPDK

Network applications based on the linux kernel networking has been observed to have many tech issues. Alibaba implemented a user space TCP/IP stack on top of DPDK which can implement all kinds of high throughput and scabality networking nodes including routing, switching, NAT, ACL, QoS, L4/7 proxy, vxlan gateway and traffic cleaning.

Chen Donghua

Senior Systems Engineer,
ZTE
Donghua received his bachelor's degree in wireless from Southeast University in 2000. He joined ZTE in 2003 and was mainly engaged in Core Network PS development. After that, he has been a system engineer focusing on PS optimization and fast forwarding.

DPDK Fast Forwarding for Virtual EPC

In this talk, the speaker will share his experience on how to use DPDK features (mbuf, ring, huge page, pmd) to get huge performance improvement on vPEC, and will also talk about expectations on DPDK for further enhancement of security and reliability.

Pravin Kantak

Product Marketing Manager,
Intel
Mr. Kantak is responsible for directing Intel’s investment in Data Plane Development Kit (DPDK) as part of Intel’s Communication Infrastructure Software strategy. He oversees Intel’s DPDK roadmap, community engagement, and eco-system engagement effort. He brought DPDK open source community face to face in the form DPDK Summit events worldwide.

Pravin has 19 years of industry experience in building, marketing, and selling communication software products. He has undergraduate degree in Computer Science and Engineering from Dr. Babasaheb Ambedkar Marathawada University, India and MBA from Santa Clara University, USA.

Closing Remarks

Pravin brings the summit to a close by reviewing the DPDK Summit outcome, soliciting feedback on the summit, and discussing increased involvement in the DPDK community.

DPDK Summit San Francisco 2014 - September 8, 2014

The first DPDK Summit brought the DPDK open source community together at the San Francisco Marriott Marquis Hotel with the vision to share DPDK usage and implementation; to hear from DPDK developers, contributors, and users; and to build the DPDK community.

Jim St. Leger

Software Product Line Manager,
Intel
Jim St. Leger is a software product line manager in Intel’s Network Platforms Group.  His work encompasses software that accelerates network traffic and applications, focusing on improving the speed and efficiency of packet-based networks through the use of multicore processors and software technologies. Jim is part of the team that started the Data Plane Development Kit (DPDK.)  He is also involved in Hyperscan, a regex matching engine (01.org/hyperscan), and the Fast Data (FD.io) packet processing open source projects.

DPDK Summit Kick-off

Jim starts off the summit by discussing the objectives, DPDK history, the community, and its contributors.

Jun Xu

Principal Engineer,
Futurewei Technologies, Inc.
He has worked in the networking industry for over 15 years, joining Futurewei in 2011. He lives in Cupertino, California. His past experiences cover areas in BRAS, MPLS, QOS, Data Plane, new platform bring-up, open service platform, network virtualization, network operating system and Network Policies. He started to work on DPDK from the first release. June received a master’s degree in Software Engineering from USC.

Is It Time to Revisit the IP Stack in the Linux Kernel and KVM?

We might take too many things for granted, like the Linux kernel providing a TCP/IP stack since its inception, whereas UNIX did not. Fast forward to today’s world with virtualization, where most hypervisors derived from modern OSs also supply an IP stack. Should we take out the IP stack to let the OS and hypervisor focus on their main tasks, including process scheduling, resource management, and virtualization, or let them be the monolithic piece for all these elements?

László Vadkerti

Lead Software Developer,
Ericsson
László graduated in 1998 in the field of Electrical Engineering. During the 14 years he spent with Ericsson worldwide (Hungary, Brazil, Sweden, India, USA), he was involved in the development cycles of several products and different technologies including Fixed-line Networks (Public Switching Telephony Networks), Ericsson's ATM switch (AXD), Mobile Media Gateway and IP Routers. He has been working in the packet processing area for 10 years now, going through different architectures, like special purpose Network Processors, home grown packet processing ASICs and now packet processing on x86 (2 years).


András Kovács

Lead Software Developer,
Ericsson
András graduated in 1998 in the field of Electrical Engineering. During the 14 years he spent with Ericsson worldwide (Hungary, Brasil, Sweden, Canada, Australia, USA), he was involved in the development cycles of several products and different technologies including Fixed-line Networks (Public Switching Telephony Networks), Ericsson's ATM switch (AXD), and IP Routers. He has been working in the packet processing area for 5 years now, going through different architectures, home grown packet processing ASICs, Network Processors and now packet processing on x86 (1.5 years).

Multi-Socket Ferrari for NFV

This presentation describes an approach to support latency-sensitive applications by diving into best practices in augmenting DPDK to deliver lower jitter and high availability in addition to higher packet throughputs. We will review Enhanced Memory Management techniques and multi-process enhancements to the DPDK library foundation.We also will describe our experience and optimizations in using DPDK with a XEN Hypervisor including the addition of NUMA awareness.

Gabriel Silva

Program Manager Windows Server Networking,
Microsoft
Gabriel Silva is a Program Manager in the Datacenter Networking Platform team. He is responsible for networking offloads, such as RSS, VMQ and SR-IOV, and the TCP/IP stack, working in this space for over 2 years with Microsoft. Previous features include vRSS (virtual receive side scaling), Test-NetConnection (TNC), and various improvements to the TCP stack for datacenter networking.

Lightning Fast I/O for Windows Server v.Next with PacketDirect

Microsoft operates some of the world’s largest data centers, such as Bing, Office365, Xbox Live, and Azure, to name a few. One of the key fundamentals enabling Microsoft to operate efficiently at such hyper scale is NIC performance. This talk addresses how to drive up packet processing performance for the network functions running in their data centers.

Yoshihiro Nakajima

Researcher,
NTT Network Innovation Laboratories
He received the B.S. degree in information science, the M.E. degree in engineering, and the Ph.D. degree in computer science from the University of Tsukuba, in 2003, 2005, and 2008, respectively. He was a research fellow of the Japan Society for the Promotion of Science from 2005 to 2008. He joined NTT Network Innovation Laboratories in 2008 and studied high-performance stream processing systems and multi-layer network provisioning / control systems. He is currently studying software-defined networking and high performance software-based packet processing technologies. He is a member of the ACM and IEEE.

A High-Performance vSwitch of the User, by the User, for the User

A high-performance software switch is a key component for next-generation telecom infrastructure, especially for NFV and SDN. NTT Laboratories developed a high-performance and highly-scalable SDN/OpenFlow software switch, called Lagopus, that leverages state-of-the-art server and software technologies, including Intel® processors, Intel® Ethernet Controllers, and the DPDK.

Venky Venkatesan

DPDK Architect,
Intel
Venky Venkatesan is a Senior Principal Engineer in the Network Platforms Group at Intel, where he guides the overall effort to make networking infrastructure functions run efficiently on Intel Platforms. His current area of focus is the Data Plane Development Kit, which is a project he has guided within Intel since its inception. Venky has been with Intel for 20 years and holds a Bachelor’s in Computer Engineering from the University of Mumbai, India.

Application Performance Tuning and Future Optimizations in DPDK

In this session, one of the original authors of the DPDK library will share insight into how to best use the available tools and library hooks when looking to optimize system packet performance. The session will also provide insight into concepts under consideration to facilitate discussion and prioritization feedback into the future planning process.

Bhavesh Davda

Senior Staff Engineer,
VMware
Bhavesh Davda is a Senior Staff Engineer in the CTO Office at VMware. He is focused on enabling virtualization of real-time applications, low latency applications, high rate packet processing, high performance computing, SR-IOV and RDMA, which have historically been challenging to virtualize. He is currently actively working on NFV, and is part of the ETSI NFV Infrastructure working group editorial team helping define the industry standard for NFV hypervisors.


Rashmin Patel

DPDK Virtualization Engineer,
Intel
Rashmin is a Software Engineer doing research and prototyping on DPDK for virtualized and SR-IOV network interfaces and secure Host-VM and VM-VM data path design to support NFV requirements on Intel Architecture. Rashmin received his B. Tech. Computer Engineering from India and Master of Computer Science (focused on software engineering and security) from ASU and then started working at Chandler campus in 2011.

DPDK in a Virtual World

The usage of virtualized DPDK applications has increased tremendously. This session will review how the DPDK APIs take advantage of platform technologies like SR-IOV, direct device assignment (VT-d) and para-virtual as well as emulated devices offered by the underlying platform to achieve higher packet throughput at predictable latency. The session will primarily focus on the virtualization options offered by DPDK for the VMware ESXi Hypervisor environment. The session will conclude by sharing the future vision and commitment to enhance the API even further to enable community developers and end users to get most out of underlying Intel Architecture and Hypervisor target.

Thomas Monjalon

Packet Processing Engineer and DPDK.org Maintainer,
6WIND
Thomas has worked on Linux projects since 2005 with a background in networking, embedded and open source technologies. As a packet processing engineer at 6WIND, he is focused on DPDK.org with the objective to make the DPDK community focused with high quality. Thomas assists customers with the integration of DPDK and 6WINDGate packet processing software for high performance Linux networking.

High-Performance Networking Leveraging the DPDK and the Growing Community

High-performance networking stacks can be designed using the DPDK and packet processing software. This presentation covers the development of high-performance applications with examples for IPsec, TCP, virtual switching, and virtual networking functions for NFV.

Tim O'Driscoll

Software Engineering Manager for DPDK,
Intel
Tim is a software engineering and program manager overseeing the DPDK project and development activities at Intel. He has a strong telecom background having spent his career working at Ericsson and Motorola in positions including engineering and system and solution architecture before coming to Intel. He’s originally from Ireland and has traveled extensively including multiple posts around the globe.

Closing Remarks

Tim brings the summit to a close by reviewing the DPDK open source journey, soliciting feedback on the summit, and discussing increased involvement in the DPDK community.