Hongjun Ni has been focusing on Cloud Networking and Network Security. He is FD.io VPP Maintainer, UDPI Project Lead, Sweetcomb Project Lead and NSH_SFC Project Lead. He has fifteen years' rich experience on Cloud Networking, Network Security, SmartNIC and Wireless. He has given 20... Read More →
Most IPSec VPN gateways rely either on the Linux Kernel IPSec stack or on a proprietary implementation of the IPSec data plane (AH/ESP). FD.io VPP incorporates a high-performance implementation of IPSec and can leverage both hardware and software DPDK backends for increased performance.
In this talk we will describe a containerised IPSec implementation based on the FD.io VPP IPSec data plane and the StrongSwan IKE control plane - using the Ligato VPP Agent as the interface between the two.
We will also present the results of performance testing of our implementation in a testbed using an enhanced version of the FD.io CSIT codebase.
TLS is fast becoming a critical networking protocol in enterprise and cloud. Open source projects such as Nginx, HAProxy or Envoy need to implement their own socket-based SSL layer as TLS is a must-have feature. It causes so much duplicate work in industry. VPP fundamentally changes this programming model as it moves the TLS implementation from the applications to the socket. Our presentation firstly introduce TLS design in VPP, and then introduce the work we did to apply VPP stack to Nginx in TLS. We have integrated Nginx with both VPP communication library(VCL) and VPP LD_PRELOAD. VCL layer is application TLS aware solution, and LD_PRELOAD is to provide to the total transparent solution to Nginx. Our presentation summarizes the work and provide the guidance how to use VPP user space for application.
Although packet forwarding with VPP and DPDK can now scale to tens of millions of packets per second per core, lack of alternatives to kernel-based sockets means that applications cannot take full advantage of this speed. To fill this gap, over the past years VPP was extended with a high-performance host stack designed to allow containerized and host applications to communicate at high speed both intra and inter-host.
In this talk, we will provide an update of the recent progress of the host stack’s major components: session layer, transport protocols, communications library (VCL) and discuss how existing and new applications could leverage the benefits of this new architecture.
\Sweetcomb is a new open source project, aiming to provide an unified management interface to configure and streaming telemetry Networt Elements. It can be leveraged in Edge Networking and Cloud, such as SD-WAN, 5G, IoT, etc. There are 12 industry founders, including Intel, Cisco, Huawei, Tencent, Alibaba, China Mobile etc. It published first official release this year. https://wiki.fd.io/view/Sweetcomb This paper will introduce Sweetcomb project and contains key elements: 1) Provides gRPC, NETCONF and RESTCONF northbound interfaces to integrate with ONAP, K8s, ODL and ONOS. 2) Supports High Availability and real-time streaming telemetry in Containers. 3) Supports IETF Yang models for Telecom and OpenConfig Yang models for Cloud. 4) Supports different Data Planes based on VPP, DPDK and Linux Kernel.
The fdio hicn project is a high performance software implementation of the hybrid information-centric networking architecture. hICN is an IPv6 implementation of ICN, as defined by Van Jacobson's CCNx, which brings location-independent communication in the Internet protocol. The project covers both the network and the transport stack and provides a VPP plugin implementation of the network stack as well as a portable implementation for the OS of most common clients such as Windows, macOS, iOS, Android, GNU/Linux. The transport stack is implemented as a C++14 library supporting reliable transport and also real-time transport such as audio/video. The project also provides applications natively using hICN transport such as an MPEG-DASH video player, HTTP and RTP support.
This talk gives an overview of the Ligato VPP Agent, an open-source management/control plane agent for VPP-based CNFs (). The VPP Agent allows an operator to deploy and manage a VPP-based CNF as a regular cloud-native app in a Kubernetes cluster. Developers can use the Ligato framework to extend the Agent’s functionality with custom plugins to create cloud-native networking applications, such as the NSM Agent or Contiv-VPP, a K8s CNI plugin. This talk will describe the core VPP Agent infrastructure components, such as the KVScheduler or various kv- store and message bus adapters, and give an overview of plugins used to manage/control various VPP subsystems, such as interfaces, routes, VRFs, ACLs, NAT or IPSec.
VPP is the data plane in the reference Network Service Mesh (NSM) architecture. In this talk, we cover how VPP is being levered in the reference architecture and the implications this has for both NSM and VPP.
The following topics will be covered:
* The benefits and drawbacks of using VPP in the reference data plane * What worked really well? * What challenges are present? * What doesn't work?
Finally, we cover additional work being done in the NSM community to make it easy to implement Cloud-native Network Functions (CNFs) that leverage VPP friendly standards such as memif.
If time allows, this talk also comes with a live coding demo by Frederick.
The high-performance software data plane provided by FD.io VPP makes application and host networking fast and very efficient. Kubernetes makes container workloads easy to manage, operate and network, all at massive scale. It is rational to integrate the two, to see how the best of both worlds perform together and how they can drive high performance communication in scaled-out and scaled-up cloud-native micro-services deployments.
This talk details how FD.io VPP is realising the vision of terabit networking on COTS servers, and how it integrates into the K8s (eco-)system through Network Service Mesh (NSM) and other Open-Source projects. It refers to Open-Source Benchmarking Lab data (FD.io CSIT project) for checks against what's possible on currently shipping COTS servers' hardware.
TRex is an open source, low cost, stateful and stateless traffic generator fueled by DPDK. It generates L4-7 traffic based on pre-processing and smart replay of real traffic templates . TRex amplifies both client and server side traffic and can scale up to 200Gb/sec with one UCS. TRex Stateless functionality includes support for multiple streams, the ability to change any packet field and provides per stream statistics, latency and jitter. Advanced Stateful functionality includes support for emulating L7 traffic with fully-featured scalable TCP layer.
Ok, now that I have your attention, let's correct that title. Containers are the future. But most aren't deploying them yet. That means the other 2/3 of the FD.io deployment diagram - bare metal and VM - are still very important ways to take advantage of FD.io, and specifically, Vector Packet Processing (VPP). In this presentation, we'll walk through two deployed use cases - one on premises, and one in the cloud - where FD.io is bringing performance gains - and price performance gains - to bear right now.
This talk with cover how you can recreate Openstack and K8s clusters with a VPP dataplane on the public baremetal cloud Packet with an API key and a few commands. From there we'll discuss a few use cases, benchmark results and how we are collaborating with other groups and projects to accelerate the adoption of cloud native networking for the NFV world.
Taylor is a CNF WG Co-Chair and Co-Owner of the software cooperative Vulk Coop residing in Austin, TX. Taylor has been an open source advocate and developer since the mid-90s with experience in multiple industries including Telecommunications, Finance, and Healthcare. He was an early... Read More →
