From: Ghanshyam Mann <gmann@ghanshyammann.com> Date: 2022-04-26 01:51:33 To: niujie <niujie@chinamobile.com> Cc: openstack-discuss <openstack-discuss@lists.openstack.org>,sunny <sunny@openstack.org>,'Horace Li' <horace@openstack.org>,"huang.shuquan" <huang.shuquan@99cloud.net>,"gong.yongsheng" <gong.yongsheng@99cloud.net>,"shane.wang" <shane.wang@intel.com>,"jian-feng.ding" <jian-feng.ding@intel.com>,wangshengyjy <wangshengyjy@chinamobile.com>,yuzhiqiang <yuzhiqiang@chinamobile.com>,zhangxiaoguang <zhangxiaoguang@chinamobile.com>,xujianwl <xujianwl@chinamobile.com> Subject: Re: [all] Resend New CFN(Computing Force Network) SIG Proposal>Thanks, Niu for the proposal and sorry for the delay in response.
I have raised this proposal to TC members and asking to check it. Overall proposal seems interesting to me but few initial queries inline.
---- On Wed, 13 Apr 2022 00:34:30 -0500 niujie <niujie@chinamobile.com> wrote ----
Hi all I sent an email yesterday about NewCFN(Computing Force Network) SIG Proposal, I tried to recall it because therewas a typo in email address, then I get recall failed msg, so I assume the emailwas sent out successfully, and plan to keep it as it was. But I found that the “recall” actionwas logged in pipermail, it might cause misunderstanding, we are sure about proposefor a new SIG, so I’m sending this again, sorry for the email flood J
I'm from China Mobile, China Mobile is recently working onbuild a new information infrastructure focusing on connectivity, computingpower, and capabilities, this new information infrastructure is calledComputing Force Network, we think OpenStack community which gathers globalwisdom together is a perfect platform to discuss topics like CFN, so we areproposing to create a new SIG for CFN(Computing Force Network). Below is CFNbrief introduction and initial SIG scope. With the flourish of new business scenarios such as hybridcloud, multi-cloud, AI, big data processing, edge computing, building a newinformation infrastructure based on multiple key technologies that convergedcloud and network, will better support global digital transformation. This newinfrastructure is not only relates to cloud, it is getting more and moreconnected with network, and at the same time, we also need to consider how toconverge multiple technologies like AI, Blockchain, big data, security to providethis all-in-one service. Computing Force Network(CFN) is a new informationinfrastructure that based on network, focused on computing, deeply convergedArtificial intelligence, Block chain, Cloud, Data, Network, Edge computing, Endapplication, Security(ABCDNETS), providing all-in-one services. Xiaodong Duan, Vice president of China Mobile ResearchInstitute, introduced the vision and architecture of Computing Force Network in2021 November OpenInfra Live Keynotes by his presentation Connection +Computing + Capability Opens a New Era of Digital Infrastructure, heproposed the new era of CFN. We are expecting to work with OpenStack on how to buildthis new information infrastructure, and how to promote the development andimplementation of next generation infrastructure, achieve ubiquitous computingforce, computing & network convergence, intelligence orchestration,all-in-one service. Then computing force will become common utilities likewater and electric step by step, computing force will be ready for access uponuse and connected by single entry point. The above vision of CFN , from technical perspective, willmainly focus on unified management and orchestration of computing + networkintegrated system, computing and network deeply converged in architecture, formand protocols aspect, bringing potential changes to OpenStack components. CFNis aiming to achieve seamlessly migration of any application between anyheterogeneous platforms, it's a challenge for the industry currently, we feelthat in pursuit of CFN could potentially contributes to the development andevolution of OpenStack.
Yes, it will require changes to OpenStack components but we will see based on the exact use case and OpenStack component scope. Is this include the application migration tooling in OpenStack?
In this CFN SIG, we will mainly focus on discussing how tobuild the new information infrastructure of CFN, related key technologies, andwhat's the impact on OpenStack brought by the network & could convergencetrend , the topics are including but not limited to: 1, Acomputing basement for unified management of container, VM and Bare Metal 2,Computing infrastructure which eliminated the difference between heterogeneoushardware 3,Measurement criteria and scheduling scheme based on unified computinginfrastructure 4,Network solutions for SDN integrating smart NIC for data center 5,Unified orchestration & management for "network + cloud", and"cloud + edge + end" integrated scheduling solution We will have regular meetings to investigate and discussbusiness scenarios, development trend, technical scheme, release technicaldocuments, technical proposal and requirements for OpenStack Projects, andpropose new project when necessary. We will also collaborate with other open source projectslike LFN, CNCF, LFE, to have a consistent plan across communities, and alignwith global standardization organization like ETSI, 3GPP, IETF, to promote CFNrelated technical scheme become the standard in industry. If you have any thoughts, interests, questions,requirements, we can discuss by this mailing list.
Thanks for the detailed information about the SIG scope. From the above, I understood that it will not be just changed to the OpenStack existing component but also new source code components also, do you have such list/proposal for a new component or you would like to continue discussing it and based on that you will get to know. How you are thinking about their (new component if any) releases like a coordinated release with OpenStack or independent. If coordinated then it is more than SIG scope and might be good to add a new project.
I think It mostly likes an auto scheduling system in a multi-level cloud fabric (multiple data center clouds, multiple edge clouds, multiple far edge clouds) platform. The main process of the system can be described as: first, to collect the resource usage in each cloud node which can be assumed as an openstack deployment. the resource usage can be provided by the openstack placement compomnet, second, to collect lataency, qos or bandwidth of the network resources among the fabric. third, when application deployment needs come in, the fabric should compute the optimal placement strategy and do the deployment. So we will need to create a new project for it.
By seeing the scope of this proposal (which seems very wider), I think it is not required to answer all of them now. Overall I am ok to start it as SIG and based on discussion/progress evaluation we will get to know more about new components, requirements etc and then we can change it from SIG to a new project under OpenStack or other governance (based on the core/requirement/use case it produces).
-gmann
Any suggestions are welcomed, and we are really hoping tohear from anyone, and work with you.
Jie Niu China Mobile
yongsheng gong 99cloud