Cloud Native Computing Foundation Technical Oversight Committee, 9 Mar 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: CNCF TOC Sandbox Annual Reviews 2021-03-09

Description

CNCF TOC Sandbox Annual Reviews 2021-03-09

A

All right all right so.

B

We are recording.

A

Thank you much keebert. I think amy sent round a nice link with all the uh an email with all the links to the projects. um Yes, so anyone got any comments or queries or questions about cubert.

A

Personally, I thought we looked in very good shape.

C

I think it's a good project.

D

I would echo that I think it's actually becoming more relevant as the platform matures as well. As far as you know, modern modernization for migration to kubernetes is a is kind of a backfill.

A

Yeah and they're, getting like really good metrics and adoption numbers and and being used by quite a lot of providers. So hopefully that'll help spread the um maintain a burden across a few more organizations in the future.

A

So yeah shall we um should we do the same things we do in uh sandbox reviews and just use chat for kind of votes for um accepting the cuba annual review.

D

Do we want to address chris's comment about it being very red hot? Is that some feedback we want to provide to the project going forward that we'd like to see a more diverse set of uh contributors.

E

I mean, I think, that's we. I think we want to see that in every project. I don't know if that we just state that, but you could you could respond to it. I was just making a public comment based on the developer stats that uh it does seem a little bit red hat heavy, but for where it's at it seems. Okay,.

A

And they they pointed that out themselves.

E

I mean they said.

A

The current owners are all red hat and they did say later on that they think it's in a healthy, growing status and they're working to employ a more diverse group of maintainers. So I think they're yup, aware of that.

A

All right I put in the chat, vote cube bert. That's basically, if we vote plus one if you are happy with the annual review.

A

A

The next one is kader, I never know how to say it kedarkeda whatever um now. I I think I said the same thing when they came into the cntf. I think they're great. I think it's really useful. It's got loads of people using it.

A

It doesn't really seem like it would have an independent life outside of kubernetes and I don't think, that's any fault the project. This is no in any way criticism project, but I think it relates to some of the conversations we have about like communicating about projects and like for an end user. They couldn't it wouldn't be meaningful to say. Oh yeah use cada unless they were using kubernetes and then.

A

I almost feel like we need some kind of project groupings if you're using kubernetes. Here are some other projects that you might want to consider alongside that.

A

Doesn't really have any bearing on their annual review, but it does have bearing on the fact that they want to apply for incubation, and I don't know whether there's no reason for those to be. uh You know we could say yeah incubated projects. Graduated projects can be part of a bigger family of projects or something. But I'm.

F

A

To um explore ideas of ways of organizing these large numbers of projects in a way that will be easier to navigate. So I just wonder if anybody's got any thoughts on that.

G

Do we have any similar projects in incubation or sandbox uh similar category of the projects that we already have.

A

So one that springs to my mind is um prometheus and thanos.

H

H

No problem, I just want to understand your comment a little bit um more deeply, so it almost sounds like you're expressing a concern that they only that they're only applicable to kubernetes, and it feels to me like there's plenty of projects that are only applicable to kubernetes in the cncf.

H

So can you say a little bit more about that concern.

A

I think this may apply to other projects as well, not just cada, and I don't mean at all that this is in any way a negative reflection on cada. I just mean that from uh and from the perspective of thinking about how many graduated projects we should have and how we want to help people navigate the landscape.

A

This particularly feels like an example of something that just doesn't have a life of its own, like it can only exist as part of kubernetes or kind of related to kubernetes.

A

um And yes, it's not part of the kubernetes project, which is fine, because kubernetes is already huge and doesn't need to encompass everything and not everyone has need for cada. So I think it's fine that it could be separate. I just wonder whether it tells us something about a model where we could say almost like. Not every project has to be a top level project. Somehow just does that kind of make any kind of sense.

I

Yeah another background.

C

Is I'm looking at the undercurrent.

I

I'm looking at the current sandbox um project, I see there will be more um similar project coming, for example, tencent game. They are talking with me. They want to donate a project which is a a couple of workloads controllers. I mean operators for their game engine which is quite specific to kubernetes workloads and another similar pattern is some some some other fun. Some other company tried to try to chrome, download job hpa, to sensitive sandbox and also see, for example, open crews from alibaba. It's basically like a staple workload, management, controller and uh yeah.

I

I think they share similar pattern. uh If I understand correctly, I.

A

Completely agree, I think, open cruise was another example, and we actually had the conversation with the kubernetes steering committee about open cruise and whether you know they had any interest in it being part of kubernetes and they didn't, which is totally fair. But yeah.

E

I mean you have to remember: the incubator was killed off, so we kind of act as the release valve for a lot of these projects. Essentially, what sandbox is for for a lot of folks.

A

So I think, with my um my comment: there I've slightly derailed talking about cada itself, um but I guess I just wanted to flag up that idea of. Perhaps we should be thinking about this grouping or family or something for projects and reading the cader doc really made me think about it.

A

um But, aside from that, anyone got any comments or concerns about keda.

G

They seem to have a good community, a diverse community, which is very good yeah.

D

I

Yeah, I personally uh be supportive to kita, because I still they have a very good contribution from diverse community. A lot of people are contributing components and integrations to this project, which is a good sign for this project, and so I'm I'm I'm personally positive to this project.

A

Leon um proposing or that they put at the end of their um annual review that they are planning to propose to graduate. I think they mean planning to propose to.

A

Graduates but uh yeah, so I I know um tom had reached out to me and said: would I be interested in sponsoring them to incubation? I was like let.

F

Me read the annual.

A

Review first before I say.

G

A

All right shall we do votes for kira unless anyone has any other points they want to.

A

A

A

And the last one is service mesh.

A

Interface, which anyone any comments or thoughts.

A

I think it you know looks good seems to have lots of momentum and it they've actually said they also plan to file for incubation soon.

C

How are users using it? I guess, is the kind of question I'd have like it's like. There are implementations, but okay are users using those implementations in order to be able to switch providers, or I mean there's not much comment about how our end users are using this or how they say it. I guess it's I mean obviously incubation. We would ask those questions directly, but there's nothing about that here, which is.

A

Yeah there is some uh comment about uh yeah, while some details for incubation appear to be focused on suitable metrics for source code, not necessarily specifications.

A

So I think they're also a bit aware that they're going to have to you know it's harder to look at adoption, metrics for a spec, but.

C

Well, yeah, except that people you'd expect users to be programming against the standard.

C

Against the spec, to some extent I mean.

E

It's you think users or actual implementations of of the spec, so things like you know all the service messages would have adherence to the spec versus necessary users. Programming against it. User is just going to go directly, install a link or d or an sdo right.

C

But if you use linker d and then you just- I mean it's only a value if you, if, if you actually are using just the smi endpoints or something otherwise you're just.

F

Going to link a d so you've.

C

F

C

The user has to actually view that the the smi conformance is giving them some kind of.

A

Value sig network we're doing something about. um uh Maybe it was more about performance and conformance, but it would be really. It would be quite nice if smi turned into a thing that you know we could say. Service meshes are conformant to or not.

E

There's some work being done in the community around that. I think lee from sig network has a kind of effort around smi conformance, but it's still like super early days from from my perspective,.

A

E

We've got this great.

A

List of ecosystem service meshes, and I suppose that doesn't actually answer the question of. Can you interchangeably swap one for another because they support smi.

B

Yeah or even like, is there any real value to smi like the fact that you can swap them? Do people swap them or do they get any value, or is this just like a tax, all service measures pay and then no end user gets any value out of that.

A

Maybe that's a question for incubation, though, rather than annual review.

C

If they're going to go for incubation.

A

C

Yeah, I I think, that's probably the best place to ask you, because that's.

A

C

That is one of the big incubation questions, but it's it. It would be nice if they had it addressed it a little bit here as well.

F

They have a line about updating the adapters for ongoing compatibility as a goal. So maybe maybe it would be nice to like document how much of the functionality of the actual implementation is covered by the smi and how much is outside. I don't know just to have a an idea.

I

F

A

Yes, I mean, certainly in terms of support from you- know: lots of different companies. There seems like there's good momentum here, so um yeah.

E

It reminds me of like maybe csi in the early days where things were a little bit messy still. I just still think they have some time to kind of build enough momentum to get somewhere useful.

A

All right, so we do votes for smi for that annual review.

A

Okay and your reviews are done hooray, so what other topics I think we've got.