Project Proposal from China Mobile
|
Kaiyuan Duan
Our team from China Mobile are preparing to create two new project proposals recently, But I am wondering is this the right way to send emails to this list if we want to have a discussion about the feasibility and the details before the real proposal. One proposal is about the Hardware TTP(Typed Table Pattern) and the other one is about GNT(General Network Topology). If this is the right way, I will give you more details about our plan. Thanks, Kaiyuan |
|
|
|
Good place to start. You might also cc discuss Ed On Thu, Jul 20, 2017 at 1:51 AM, 段凯元 <duankaiyuan@...> wrote:
|
|
|
|
Robert Varga
Hello Kaiyauan,
On 20/07/17 10:51, 段凯元 wrote: One proposal is about the Hardware TTP(Typed Table Pattern) and theI think it is worth drafting project proposals, at least to support the discussions of what is being proposed. Thanks, Robert |
|
|
|
Kaiyuan Duan
Thanks for your reply, I will first share our ideas and the details about the two proposals here: 1, Hardware TTP Our team have defined a standard TTP(Typed Table Pattern) pipeline ，which is adaptive for Neutron at the Northbound, for the hardware switch. We use OpenFlow on hardware switches to manage the communication for Layer 2&Layer 3, including the communication between physical servers, physical server and VM, server and vRouters. It can also support the security group policy management, QoS and some other basic function. I know there is already a project related to TTP in the community, but it has not updated and code contribution for a long time and its orientation is also unclear, so we hope to lead a new project. It is ok if you prefer a project integration, but we need to solve the collision before that. 2, GNT(General network topology) The goal of the project is to provide a unified topological layer, which can monitor different kinds of devices(ovsdb, hardwareVTEP, netconf) and computing resources(physical or virtual machine). This layer will shield the difference of the southbound protocals, make other business modules easily get the overall topo without adaption for different protocals. Besides the function to extend logical topo through the code itself, we have defined the yang model to provide the possibilty that let business modules for other southbound protocals get data from or write data to the yang model tree. In addition, this project refers to Neutron of the northbound configuration to analyze the distribution of different kinds of network resources. The existing topology module is focus on the OpenFlow protocal and no design for the shield of other protocal resources. Thus, we hope the our project, GNT, can extend the topology of logical network and reduce the coupling between modules. We still have some unclear parts and we hope you can give us some advices. Regards, Kaiyuan ----邮件原文---- |
|
|
|
Abhijit Kumbhare
Regarding: "I know there is already a project related to TTP in the community, but it has not updated and code contribution for a long time and its orientation is also unclear, so we hope to lead a new project. It is ok if you prefer a project integration, but we need to solve the collision before that." I am sure folks on the TTP project will welcome your contribution to the project directly - especially since you have your own TTP pipeline that you would like to use. Your TTP could be a good real world use case for the project. And eventually the new contributors can become committers on the project. (By Folks for the TTP project I mostly mean Colin - as he is the active committer/contributor. But as someone listed as a TTP committer on the TTP wiki & who would like the project to do well - I would more than welcome your contribution). On Thu, Jul 20, 2017 at 9:37 PM, 段凯元 <duankaiyuan@...> wrote:
|
|
|
|
Some recommendations: 1) Reach out to the existing TPP project first. You can have a separate project if you want to (and it may even be the right play), but its useful to talk to adjacent projects first, if for no other reason than they may have pointers as to where the hard parts are in your problem space :) 2) Take a look at the yang topology model. If it meets your needs, we have various places using it already :) Ed On Thu, Jul 20, 2017 at 9:37 PM, 段凯元 <duankaiyuan@...> wrote:
|
|
|