Cloud Native Computing Foundation KubeCon + CloudNativeCon North America 2020 - Virtual, 4 Dec 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Empowering Cloud Native Networking with Arm Ecosystem - Trevor Tao & HanYu Ding, Arm

Description

Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon Europe 2021 Virtual from May 4–7, 2021. Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.

Empowering Cloud Native Networking with Arm Ecosystem - Trevor Tao & HanYu Ding, Arm

Arm ecosystem is becoming much more popular in cloud native applications than ever before with its increasing wide use. Arm devotes to be a cloud native vendor and puts much resources to enable related projects on its platform. In the presentation, the following key takeaways would be given but not limit to, which would exhibits our work on cloud native projects, especially for networking: 1. The reference cloud native reference stack, use cases and its supporting technologies on arm; 2. The enabled K8s networking solutions and service mesh for arm64 platform: Cilium, OVN-k8s, Envoy etc; 3. Senior features when meeting multi-interface requirements with SRIOV/DPDK by Multus/SRIOV CNI; 4. Performance evaluation and bottleneck fixing with the typical tunnel solutions(IPIP) used by most CNIs on arm with respect to x86; 5. Our exact contributions to the related OSS projects and future work

https://sched.co/ekA0

A

Hello everyone good afternoon today my topic is empowering cloud native networking with arm ecosystem and my partner hanyu ding from china. Mobile would also give a talk in this session. Now I will start this session. Thank you.

A

Firstly, I'd like to give out the agenda for today and the first day is a introduction for our brief introduction for our today's topic and the second one would be reference cloud native networking stack on arm. The third one would be high performance container networking interfaces for canonical networking on and the first one would be the multi interface requirement. This is rv support on the fifth one, with the I we would like to give out some performance test results for various container interfaces on arm and the sixth one would be.

A

Six item would be cloud native, networking use cases and deployment arm and our the last one would be our future work.

A

Firstly, I would like to talk about what is cloud native networking currently to computing means to use containerized open source software stack where each part of the app is packaged into its own container dynamically orchestrated, so that each part is actively scheduled and managed to optimize your source utilization cloud native networking is an approach to provide the networking environment for building and running applications that exploit the advantages of the cloud computing delivery model, so high performance, flexible in a function.

A

Reach containing networking technology is the key to the success of canal native computing for various cni's. An emergency today is to support the canal native networking.

A

Cni is a cn's canonical computing foundation project consists of a specification libraries for writing plugins to configure lateral interfaces in linux containers, along with a number of number of supported, plugins and the cni as follows. That kubernetes follows the kubernetes networking model, our native computing with um um originally joined the linux foundation 10 years ago and joined the cncf as a golden member in february 2019, and upgraded to a platinum member financing, the member on november 2019 and um is already a key player in teleconcentric organizations.

A

Such as tel kong telecom, infra project, dbdk, fido and english foundation's aquarium project and the arm provided arm nearers the foundation for the 5g cloud to edge infrastructure and the arm is actively engaged with the cloud native software ecosystem to deliver optimized stacks. So devops teams can port and run codes and applications on arm-based solutions.

A

I'm provided neo-words platform to support the cloud-to-edge cloud-to-edge computing. For example. The um currently the arm on one platform provides a very high performance scale, high performance, high performance support for the cloud to edge computing and they for the next generation and provide an online platform, provide 60 performance uplift from the original a72 uh arm course, and for the next generation for 2021 next generation n2 platform.

A

We named it as perseus per cells and performance fixed, a semiconductor technology from 7 7, nanometer, q5 nanometer, and for the um neoverse until platform we provided the cloud to edge efficiencies for the cloud data centers. We pride 32 to 192 cores, which has a very high rather high performance, but with very high high core count: high power consumption for cloud cpus for the middle middle clock for the middle core for the add use case and the faculty use cases.

A

We provide the middle core count with 12 to 36 cars and with the median tdp median power, consumption pulse consumption for the 5g and to edge and something like switches and symanix cpus for the no end, node and edge devices for 5g or other edge devices.

A

Will provide very low core count, low power get for the gateway and route cpus.

A

So it's a serious platform, scalability scale views the performance requirements for each use, cases and actually cloud workloads deserve dedicated costs for the hyper-threading one. The data center and the threat performance will win the cloud with more and more more and more high performance cross high performance call half performance threats.

A

Now, I'm had in name I'm enabling a frictionless cloud native developer experience from a wells open source pro open source software project, such as from the workloads such as say.

A

They now would like to give the third one set apart is a reference cloud native networking stack on arm in this reference cloud native networking stack. If I be able to give, I briefly use the hardware acceleration from semanic or integrated accelerators is demonic and from there we will use the operating system network. We will provide the real-time linux distribution, retardants and srv and dptk support from the oxidation layer from the orchestration layer.

A

We would provide the kubernetes orchestration for the containers and containerized installers and restock and other things like that from the networking into software. We would provide vpp obs and linux system, networking with ebpf support and for the container working the nano catechol serium, this ebpf and cantilevp vstbdk support and ovm kubernetes is always support.

A

We would also be provided, and as a high level, we would provide the kubernetes networking abstraction, which provides load balance policy, high availability, dns lifecycle, external ips and securities and for all the service providers, would be like something like cncf service mesh package management, ci, cd and others. We would like to support the 5g network, functions, mac and other web-based micro services.

A

So we'd like to talk about the core of this cloud native networking stack is to the supporting network technologies for layer 2. We would use linux ethernet driver bridge, tpdk, virtual, switching or things like that and for layer 3. It would be ib tables and for layer 3. There will also be overlay tunnel for ipip vx line and the vagi and other ib tunnels other ip based tunnels for air 2 and 3.

A

For the l2 3 technologies, the most important one will be the kernel jit, which is ebpf ebpf buckley package, filter, enhanced each provider programs, maps registers help functions all things like that to support the high performance or service oriented nowhere, support in the kernel and from the l7. We would apply the http proxy for the unwi based online based http proxy, and this is the reference cloud native networking stack provided provided on arm.

A

Now would like to. We would like to talk something about at the high performance container. Networking interfaces for the container.

B

A

Cloud native networking or arm.

A

Actually, the high performance container interface is available for arm is. uh There is a lot of a lot of support for the high performance container network interfaces on arm, and we would like to here. We would like to give a brief conduction from the our iec reference stack, iec reference, which is from ocarina ic blueprint, which is provided and integrated into ic repo, and I had to give out the links you could. If you are interested in this one, you could check out this cni's.

A

The first one would be calcul is a pure ipv networking based and there is a high level policy network policy management by ib tables. It provides good scalability and it supports direct and overlay network collection and with easy, easy deployment, and the second one will be will be syrian, which is a very popular networking. Support containing networking support today, which uses linux ebpf based on network policy, the balance and the security which is believed to be this incredible performance and the air 3.

A

Networking beating house is good scalability too, and the cloud continue vp, which uses fido vvp, to provide the network collectively bidding ports and which uses native dbtk interface, support for physic and for oem kubernetes. It is ovs ovn, controller-based, kubernetes, networking solution, rather good performance with ovs inherited and uses oem, logical switches and routers to collect parts for outside access for the mark for the srv fit physical interface support. We provide monitors with srv synes, which provided the right direct physical interfaces for pfbfs support for ports and which is high-performance visa direct.

A

This direct linux kernel is not a driver or dpdk pmd driver and it usually co-worked with other cni such as financial, calico, biomartins and other glucan eyes, and the last one would be the widely used nano used, which is a easiest deployment for simple kubernetes networking which uses linux nanobridge for port communication and overlay based communication for in-house access.

A

So if you are interested in all these things, as you could find them on, um you could have anything in the iec repo in ocarina.

A

Here I would like to give a brief uh give a brief introduction about the status of shin eyes on arm platform and our contributions, which is only a small, very small part of it. um Now I would like to talk something about the mod interface requirement with our v and symmetric support on arm.

A

Here is a multiple interfaces support by srv cni. Usually, we provide a legacy, a legacy for nano and chemical plug-in for the all the ports as a default c9 and optionally.

A

The user wants to use up high performance interfaces in their port, so we provide the srv thing as a plugin single plugin to provide a secondary internet to private second interfaces interface, and the second interface would be the physical interface provided by the physical nic, which is maybe a vf or via pf of a physical nic. Here we provide this srv support.

A

Pvf support to the port with a smart nik, which is from the broadcom, which is from.com here, is the right part, provides someday resource configuration for the resource configuration for the uh physical for the vf mix, for example, for the srv interfaces from the uh broadcom. From this broadcom fifth manic.

A

We provided the drivers, you know bnxd internet driver and we provided description about this driver and about these things, so the system will discover all the remain all the related virtual functions in the virtual function interfaces here and if we want to provide the dpdk device drivers, we would like to use the dbtk drivers, which qbt drivers, which we have our pci here is a physical user physics clinic by intel here is a network attachment the definition for the resource srv.net we reach it would be used in a port description and provide the ipam support, ipms for the ip address or location for the port, for example, if you want to provide srv, srv srv interface into your port in uh besides original legacy legacy financial calculate interface, you provide you provide a descript annotations here for the srv net one and we provide the resource requests here for arm.com ps22.

A

Ps2225 srv network device, one which one and one resource so it will provide a one srv interface into the pod. With this with this with this part, here is the networking model for containers in a single host by servicing line, and we provided a srv singer by smart gps q25 with a single pf, but these two different vfs and it is collected by and different vf's bothering us in a single host for the inter-host communication with other hosts. We also could use the srv single plugin, which provided vf support.

A

We have support to the outside world with the srv interface for the second host. You here, you can see. This part can communicate with another host with the sre interface, with emp9s0 to the semanic syllabus monica vf 0, and to the outside world to the outside host.

A

So we had, we had done some performance tests for the various container interfaces on arm platform. The first one would be here. We will provide a performance test for different clients with different back end pick and support. For example, we had tested the calico with ipr tunnel and the catechol v director routing and the fernando is ipr tunnel for nano with vxlan tunnel excellent tunnel and serium with vxlan tunnel.

A

For example, we get we get the results as follows. uh The initial observation for the tcp performance over different thing eyes we got is, for example, we get to know the performance gap between cni's are also explicit when overlay tunnel is used and the calico and the nano show a little bit better performance than cdm for most mq skier, and this ipip makes line overnight tunnel import enabled as a large mtu site.

A

Throughput performance is better and when used directly routing the software performance is not significantly affected by mq size here, as the time is limited well, I would like. I would not talk too much about the results you could check that they check it check them. Detailed check the details by yourself in the latest in this slide lately and a very interesting result here, is that we had done some http performance. Benchmarking for synonyms with various backends.

A

Here is the http to host to service hdd performance, which is provided by kubernetes for kubernetes for different from different cnis 10 gigabit collection batteries and line gigabit performance result with the increase of the threat we used in the iproof test.

A

Here I would like to say something about the service mesh on arm service. Mesh service mesh is the communication layer in microservice setup setup, all requests to and from each other services go through the mesh, also known as infrastructure layer in a micro service setup, the service mesh makes communication bidding services reliable and secure for their control plane.

A

We uses the istio and which is composed of the mixer, which is for policy enforcement and telemetry checks, pilot for service communication policies, configuration gallery, control, plane, computer configuration and see title sedation, security and credential management, and the from the data plane is on y, which has a http proxy for the high performance proxy developed in c plus plus, and on.

A

Why ony provides a lot of service such as service stakes, discovery such as rate limit, such as rate limit and for the l7 field, l7 field and the advanced, node balancing and all things like that, and we provided a sto use cases for the istio booking for exam sample on arm.

A

We had deployed this still example, an example on this arm platform, with the product page red page reviews on the ports and on my work here, works as a center card to each part, which supports a micro service architecture for the access data flow for the different accesses.

A

Now hyudin from china mobile would talk. Something about would talk something about the use case which, uh in china mobile provided for the arm, prep.

B

As we all know, the cloud native technical will promote 5g services in china mobile to be more flexible and convenient. The virtual machine based orchestration is trans transformed into container and component based orchestration. The gradual, the gradual precipitation of operator capabilities will will further enhance business, creativity and flexible ability.

B

In recent years, china mobile has actively invested in the open source field of edge computing and china. Mobile has been elected to the acronym community technical committee and has and have actively lead many as cloud blueprint projects in atreno.

B

um So I will introduce the some use case in china, mobile uh china, mobile and its partners jointly initially and to be the ptl as the two adrenal uh projects, which has which have successfully released the rf3 version of the community, and the first is uh about the cloud android cloud native applications on amp server in age, and this project mainly uh provide the end-to-end android application solutions around android cloud gaming scenarios.

B

The cloud gaming is one of the key vertical industrial applications of 5g edge computing in china mobile. It mainly deploys cloud game, rendering and editing in at add node and some decoding storage. Some other functions.

B

The project will provide the android cloud native running environment under the arm architecture for mobile terminal applications at the age, and it also will reduce the difficulty of deployment and the development of applications on the s cloud and reduce the deployment cost of android. Add cloud.

B

Next, the second project is about the smart nick, the rf, the r3 version of the integrated as cloud type 5 bluepoint released for the first time the oas offload reference, implementation of the blue field network and card based on the soc architecture, and it also merged into the acronym community. R3 uh version and the first release of the um this project is based on the am soc architecture and obs.

B

Dpd3 is unloaded to the smart network card, which can enhance the support performance of edge network vpc, reduce the patched loss right and enhance the management of network card resources to save more computing resources in the future and the performance of 5g upf network elements deployed in the s-cloud data center in china.

B

Mobile can also be enhanced by realizing the uh uninstallation function of a network card, so mobile is welding to further promote the maturity, business, innovation of the cloud native computing through the active exploration and practice in the field of cloud native and computing open source, and we are willing to create a win-win cloud native computing intro system.

B

A

Okay, and can I support the enhancements on arm platform and, for example, for ci cd performs a future comparisons, and we would like to give some performance optimization of overlay tunnel communication on armed platform and further use cases and deployments about cloud native networking on arm platform, for example, arvr, 5g mac and the service mesh integration with high performance, such as cdm, proxy and further dpdk, incorporated container networking usage model and the performance evaluation of other ci for high performance connectivity.

A

Solutions on this updated processes here is some references used in today's topic uses in today's topic, and thank you all for today for attending this meeting.

A