Cloud Native Computing Foundation Network Function Working Group, 29 Mar 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: CNF WG Meeting 2021-03-29

Description

CNF WG Meeting 2021-03-29

A

Hey thanks everyone for joining I'll, put a link to the meeting minutes in the chat. We'll get started a little bit after the hour here, and thanks for joining.

A

B

Hi thanks for joining we'll get started at about five after.

B

The zoom chat should have a link to the meeting notes and you can add your name to the attendees list and any agenda.

B

A

Okay, it's five after I think you can probably get started so once again in case you're, just joining, um I drop the link to the meeting minutes in the chat feel free to add your name. This is the weekly meeting of the cnn working group. We meet every monday at 1600 utc before we jump in. Is there anything that anyone would like to add the ad to the agenda.

A

Okay, hearing nothing, um we can jump in so the first thing I wanted to talk about is the voting. So thanks everybody that's participating.

A

um Does anybody feel first of all, did anybody not receive a ballot and think that they should have.

A

Okay and second question: it: will anybody not be able to vote by tomorrow night and would need an extension on the deadline for voting.

C

Hey bill, I put the time zone next to midnight just so, people are tracking if, for whatever reason, they're cutting it to the last minute.

A

Okay um by midnight, uh ah yeah thanks um yeah and then the other thing is so once we have the co-chairs elected, we can start um the tech lead elections after that.

A

um Does anybody have any questions about the voting process so far, anything that I can help them with.

A

Okay, if you do, please feel free to reach out to me on the cncfs super happy to help just trying to make it as easy and as transparent as possible for everyone.

A

Okay, um second one: we have a look from ericsson and he previously presented uh eno, which is, I think, the external network orchestrator at the tog meeting and now he's interested in putting together a use case for kubernetes network orchestration.

A

um So I'll give it to him to just give kind of, like a quick update. uh Let's say, maybe like three to five minutes about what he's doing, uh because he wants to put together a pr and is looking for people to help out so over to you.

D

Yeah thanks bill uh good morning, good afternoon, good evening to all yeah, so we presented you know that is the external network operator in the last telecom user group.

D

Many of you might have joined that meeting as well and have gone through the proposal and the idea which we are kind of trying to float in this community, and I had a follow-up discussion with over the email with bill and tal about how we can take it further and one of the suggestions which I got um based to write a proposal in as in like uh use case.

D

uh There is a template which is available in the cnf working group repo and we just followed the same notations and nomenclature and been writing the proposal. It will be available in couple of days one or two in in in in the form of pr in the repo, but so if, if anyone would be interested to be as part of reviewer, please let me know maybe over the slack or just in this meeting I can add I'll, be happy to add them as as a reviewer and yeah. That's it.

D

I guess from my side.

A

Cool, thank you um so looking forward to that pr and the use case coming out. So if anybody wants to be added as a reviewer to the pr, please add your github handle to the meeting notes, and I can add you there so any questions about that or we can keep moving on.

A

And thank you for the people that are volunteering um to be reviewers for the pr yeah.

D

A

Yeah I'm looking forward to that.

A

Okay! Next one is the clarifying the chairs responsibility. Jayla did you say something.

B

Yeah I was just gonna ask if um I think that one's gonna be a pretty big um use case and discussion. If a um discussion item could be added to the github repo that links to the pr, so that we can um continue that conversation, there's been a lot of stuff on email and um in other slack channels, and I think it'd be good.

B

If we capture more than what what the pr is the rest of the discussions, could you create a new discussion on github for this use case that you're going to do a pr for sure sure I can do that yeah that.

D

Would be a good option? Yeah thanks.

A

Okay, cool thanks; okay, so the next one is clarifying the chairs responsibilities. So this is a pr um from taylor, essentially editing it so that it kind of clarifies that the chairs are for facilitating kind of the group's interactions.

A

And we have quite a few approvals here. I just wanted to check in this meeting any last comments with this pr. Otherwise we can merge it.

A

Okay, cool, so thanks taylor for that pr, um the next one is adding the glossary to the contributing. This is my uh or this is this is lucina's pull request. um So thanks for updating this, this is adding essentially just a link to this and the contributing documents that, as people combine, are wondering where they contribute they can add to the glossary.

A

I think this is pretty straightforward and unless there's anything else, I'm quite fine with merging this one too.

A

Okay, so thanks lucina for that.

A

um The next one is ian on the call.

A

Today, I don't see them, but um in case people had missed this one. This is about the rules. um Yeah thanks, lucina um is about kind of. How do we do?

A

How do we see what companies are associated with each other? This is out of the discussion about the voting rules too. um We have two approvals so far. I guess. Does anybody else want to like look at this or be added to the reviewers?

A

I think maybe we wait another week until a couple more people had time to look at it.

A

So does anybody want to be added as a reviewer? I can just do this right now.

E

Bill, can you have my name there.

A

Yeah, that's victor right, yep! Thank you! Okay, cool thanks.

A

A

Okay, cool, I think the.

B

Match is, is clarifying um cleaning up more than anything and moving a lot of it into one shared area.

A

um The next pull request is around having a timeline for the interested parties, just for legal departments that want to be more clear and people can give people deadlines. This is essentially a one-line pull request. Actually don't know.

A

Is this going to go because taylor, you pulled out the interested parties into a separate um into a separate file.

A

Will this go into that file.

B

That would make sense to me what that would make sense that it should go over. Of course, I think with the merge. This would probably say um something like go look at the file and then interested part. I I don't know why the um subheader hasn't updated, but yeah, okay, conflict.

F

But this is, this is just saying that the interested parties can be added at any time. I think what needs to be added is to the sentence can be added at any time to the list and you point to the list, but the, but the statement still belongs here.

B

I think it's once uh sentence up. If you go back to that conflict.

A

B

Okay um above uh each organization, yeah right where it says each organization.

B

So this is where the interested parties is and then um I guess it would go if you wanted to put it, maybe in in that paragraph right before any contributors.

F

But that is about the that is about the elections. It is not about interested parties yeah. I think what is good if I think you need a section interested parties before you have a section on elections, and it should say that you can add interested parties at any time or whatever time frame you you want to specify and where that list is.

F

This is in the election section. This only says each organization shall have one vote, that's what it is right.

A

F

The definition of what is an interested party and how you modify it uh needs a separate section by itself, just a thought.

B

Well, right now, we've we've removed that what are interested parties and everything- and I we've already said that we would like to um look at changing the whole voting as well as representation.

B

So I don't want to do a whole lot of work on this. I would say either remove it and just kill this pr for now or do something minimal like what you're doing um bill you can put.

A

It on or yeah I I get what you're saying penn college. Would you be okay if I.

F

That's fine, that's fine! Yeah.

A

Yeah, add it as is, and then add an issue that the collection process should.

E

A

A

A

There's still a conflict in.

A

A

Oh did I do this the wrong way.

F

A

A

A

A

uh I'll add a note that we should yeah file.

A

So we'll create an issue for that cool um you, and I I saw that you commented here about this.

G

Yeah, uh really it's not um I'm just saying my opinion. um Other people no doubt have other opinions about what a use case is it shouldn't say a use case should not say: eno is great eno does this, you know is marvelous. A use case should say here is an abstract problem. I have to solve so that you can show that eno is great by solving the problem in a separate document.

D

Sure yeah, I took a note of your comment in the slack channel and definitely will put along the lines. Yes for yourself. Yes,.

G

That's that's fine. I mean I guess I've got my skepticism about whether eno is right, but here's your chance to prove that you, you know, demonstrate it to me, um but I just want you to. Basically, if you write, eno is a thing that we will be using.

G

Then then you don't really get the chance to prove how great it is. That's the thing I want to see make sense. Yeah, okay,.

B

um I don't know if you heard ian, but this, that what you're talking about is uh one example of why we need a a github discussion so that we can add more comments than just the pr.

G

uh Yeah I mean I would take either, but I totally sympathize with what you're saying um I'm not gonna say: don't do it because you're doing it in academia in administrative slightly the wrong way, but you know it just might mean that the pr doesn't get accepted, but um yeah um either way.

C

Well on this topic, though, right like the use case, should be network orchestration and should be like completely agnostic, then a best practice using eno would be a proposal right, like the whole discussion versus exact pr. Is this and that, like like what use case, are we trying to fix that's a pr? How are we going to fix it? That's another pr.

G

That's where I was going with that yeah.

A

And speaking of that, we do have one use case on the table right now. This is from book, and so this is about the what happens when you need to manage the infrastructure life cycle and what happens to the cnf during that uh jeffrey? I saw that many comments nine minutes ago, um but we're also looking for more people to add some reviews to this.

A

So is anybody else interested in reviewing this and I think the process of like how this gets like approved and managed through, I think, we'll become more clear once we have some tech leads that will kind of like help shepherd this along too.

G

Can you add me to the review list? Yes and I was planning on reviewing it as well. If you add me as a reviewer, I suppose that's a reminder.

G

A

C

I think this is a use casey and that would spawn multiple best practices personally um and so yeah. I did a couple last minute. There was a few that I forgot to hit um submit my review.

C

So that's my bad bill, but I did a pretty sizable commit like I talked to book and just did some like basic, like editing and stuff, but I think there's a lot to unpack in this first one that brooke put up, I mean there's stuff around testing. I think there's best practices. We should write around testing, there's best practices around like resiliency.

C

That could come from this based on, like some of the m things, that book laid out, um there's even like best practices, and I don't think we would necessarily cover this, but just like the interface between an app team versus an infrastructure team like we talk, I know you and you and I have had a lot of conversations around um like what slas look like what is like the mutual um relationship between people building the app people hosting the app people running the app, and um so I I mean I don't know if that was book's intention, but he put quite a lot into this first use case, um and so I think that once again to the earlier comment about you know, it's like a use case.

C

You know is in itself just something that we're doing, and then one or multiple best practices could potentially you know, try to address, said use case and invokes instance. I think it would take like almost a baker's dozen to unpack everything.

G

Which is fine, I'm good with that I mean it seems to me that one of the more important things we get out of this is this is what you should expect the platform to promise you uh or you know this is what the platform should promise. You know, however, you want to whichever lens you want to view that in so which would be really useful. I mean, if you know, someone to actually declare the platform. Is this reliable and no more would be very, very helpful.

C

Sure, and in reverse, though, some of the stuff that book puts in there and he's obviously coming from the bias of being someone who you know provides a platform is these are the types of things that we need to do and we're expecting a cnf to be able to cope with them.

A

Yeah, sorry, I'm just trying to write these down in the background, um so we can get them so I'll. Add these after the meeting, because I don't think people want to see me do that.

A

um So thanks for the people that have raised their hand to um review that, um definitely super helpful. I don't know if people will have more comments about this use case today or want to have like a little. We can also it might be helpful to have some more people. Look through it and generate some more discussion before we dive more deeply into it.

A

Cool um next one is creating a pull request template. um This is an issue from lucina. I mean something we might want to consider to help out people that want to contribute, um and so the type of pr the description, the motivation contacts, and I think we can actually probably use like some of our include some of our templates too.

A

So if anybody wants to take a first stab at creating like a pull request, template looking for.

A

Volunteers- if not, I'm actually also interested in this too, so um I'll assign myself unless anybody else is dying to do this.

A

E

I can help as well.

A

Okay, thanks lucina, maybe I'll um I'll run something by you later this week. Once I have something.

A

Okay, the next one um ian. I hope you don't mind. I.

A

Numbered your bgp use case.

G

You're a terrible person. I hope I didn't break anything, that's fine. I have no problem.

A

Maybe if someone wants to just like look through it and make sure I didn't mess it up, that'd be good, uh it should be pretty straightforward, but that's when all the accidents.

A

Happen um yeah, uh the next one is so taylor started out. He broke all the um the the checks that we had, um but now fortunately they're well they're, not green. So ian won't be super happy, but he.

F

A

Draft pr, um because we've kind of we originally talked about conformance now we're switching to best practices. So if people want to like help out with this um feel free to jump into this too,.

G

um Taylor, I can't remember how it doesn't you'll have to read the documentation, but that, obviously it's a link check to check for valid links and in a in drafts it may well be that you intentionally have broken links. In particular.

G

um Sorry in one of my talking templates, you might well have broken links intentionally, you don't want them to work. I think you can put exceptions in for them by putting a comment in the markdown, but you would have to go read things, I'm afraid.

A

Yeah, I think he probably broke some, because it's changing the name of some of the files.

G

Oh yeah, that would do it.

A

Yeah um bf anybody wants to help out with this draft pr. I'm sure taylor would welcome the help too.

B

Yeah for sure I'm I'm trying to go through all of them. So if would if we, um as I'm go going through, I'm basically clicking on that branch, that rename conformance definitions, branch and then going through and updating all the docs to fix links like you're saying in some of it's going to be pointing to existing new docs, and some of it is going to require renaming stuff like the the actual best practice name, which is, we have conformance definition proposal, so that'll rename the folder and other things.

B

So I'm sure there's going to be lots of broken links and lots of wording, that's still using the other language, so yeah. Anyone that wants to help would I appreciate it.

H

So, does this mean that cnf conformance the repo is now cnf best practice.

A

No, so those are two separate repos. So if I go to c enough um conformance, this is actually a separate repo.

H

Yeah, no, I realize that, but if the is the test suite still a conformance test suite or is it a best practice? Is it performance against best practices? I guess.

A

H

B

So I I wasn't bringing it up on this. There are plans um split up. If you scroll, can you scroll down bill yep on to the readme? Until we see the actual content? No, you can keep scrolling there. We go you're right, so something that's confusing on this is um we're talking about the overall high level conformance program or the efforts to have um potential, I guess we'd, say best practices, but the overall program and efforts which should include maybe certification, but none of that's there right now.

B

So we have this big program and then we have the actual software. The test suite, which right now is focused on best practices and giving feedback. There's no certification, so we're planning on splitting these up and that would be similar to the kubernetes conformance, there's a repo and a whole conformance program. That's completely autonomous and runs separately as far as like the governance and decisions around that from the test suite.

B

So if you're, if you're, looking at the kubernetes e to e test the decisions on, what's you know, covered and stuff and that that's done completely separate from the conformance, which is a selection of the test, and so we'll do the same thing around this and split it up and then it'll it'll be a lot more clear.

B

The cnf working group, our focus right now is what are all those best practices that could be tested the tests we could test them and then a conformance program. If we move forward with. All of that would be saying, here's a way to say that you know you're you're following all these best practices, so it all come together.

B

Does that make sense yeah it does yeah.

A

Cool thanks for that taylor.

A

um And then there's a couple formatting issues from ian. I guess I know you created the issues. I guess do you know of any solutions or these like things that we should just take note of and um add to our contributing docs.

G

Well, to be honest, it's got to go in the contributing dots before we put a check in and the moment that actually gets done, I'm a bit ambivalent about that um line length one. To be honest, um the editor I'm using and I'm guessing that it's you know, visual studio code doesn't seem to like the idea of auto word wrapping things. It thinks that you should leave lines long and then it will word wrap things, but it makes diffs absolutely horrible.

G

If you do that, um I imagine a line length check is you know, 100 lines of python to actually make the check happen. So I guess we could write one in short order. I would prefer to make sure that it's not just my opinion on the subject before I actually go out and do that.

G

Because otherwise, I will be hated for making everybody word wrap their mark down, because it forever fails tests as opposed to everybody signing up to yes, I will do 80 character markdown, because I, like your ideas,.

H

um I guess question is why 80 is it an arbitrary number yeah? It was.

G

Really it's just a number that basically allows individual line diffs to work, um because that doesn't work right now and again. You're welcome to tell me it's a really bad idea, and you should never consider it again.

H

I don't think it's a bad idea. I just think 80 is a bit short. That's.

G

All but but you know, my green screen console only has 80 characters across.

H

I'm not going to comment on.

A

B

If we were just writing and sharing stuff, I'm all for doing it on an old green screen and limiting it at this point, though, for everyone else, I would say: let technology do the wrapping.

G

Well, frankly, it's it can sit there. You can pass comment on it. It you.

E

G

Of it as a very low level issue that we don't have to give any serious worry to if it basically lingers around for another month, and nobody really feels that it's compelling then we'll just close it and be done.

I

Okay, maybe we should find uh an example of a of a short and a long uh lying diff so that we can see what the readability looks like in github, because I think this is going to come down to how well does github show long lines like you have to scroll to the right, because it's uh because of github's lack of auto wrapping or uh does it allow you to show that in a really nice way and uh and that way it's becomes a non-issue at that point.

I

For for most of the users, though, it may still affect people who review things in in command line, but I think that that'll be uh that'll, be a minority of users who can build their own tools to auto wrap it, and I try to work things out.

A

Okay, um if you're really passionate about this comment on the issue, I I don't think we need to take a ton of time to discuss this.

G

Yeah yeah: this is not our day job. This is just a nice to have. That would make our reviews a bit easier. So don't really don't don't knock yourself out over it.

A

Thanks ian um and yeah jeff, I also added that um to the issue, so it's recorded. um The other thing is the web links and doc point to the github location.

A

A

This is what you want things to be more like.

A

G

um Yeah, it just needs relative relative links. Otherwise, what's going to happen here is any branch that you create will link back to the the trunk documentation the master documentation, so nobody will be able to review it as process markdown for one um uh forked, repos yeah. It just needs dealing.

E

G

But basically our links want to be relative and I don't think the link checker is clever enough to work that one out.

A

Okay, so maybe this should be added to the style.

A

A

So that's also.

A

This is a really good first issue.

A

Anybody wants to tackle that one.

A

A

Another one just related to that is also updating from master to main um from, like the github thing, I guess I haven't looked at. I haven't read into the documentation, but I'll probably do that later this week or if anybody else wants to or has experience, I'd love to hear kind of their experience, like switching other repos over.

A

What we're trying to do here is switch the uh default branch from master to main.

A

B

I'm I'm up for taking that on, it's pretty straightforward and we're trying to do that on all the other ones, like all the other uh repos.

B

Do you have a link in there to the um cncf or is it I don't know if it's cncf or lf naming.

A

uh No, I can add that to you, though,.

B

I think they have something in there.

A

Yeah, but this is the documentation from github too yeah.

A

Okay, thanks taylor.

B

Getting rid of language, that's rooted and um or wherever problematic because of its background, so master would be one of them and that's happened. I think, multiple years ago for uh github it just hadn't gone across all of cncf yeah.

A

Okay thanks taylor um fredrick, um it looks like you recently opened a discussion 15 minutes ago. Would you like to discuss.

I

Sure so the basic idea is to propose a bump in the wire. Well, actually, I should say a hello world use case.

I

This is something that I had spoken about a month ago or so here, and the idea is basically to put together a use case where we have an advanced use case, which ian gave an advanced use case that we can focus on, but also to have a very simple use case that we can ask the question that, as we're solving problems for the advanced use cases, are we over complicating best practices for a simple use case and a bump in the wire firewall is a possible easy use case.

I

We can focus on that that basically uh serves a specific uh that that specific role and the idea was to basically have it set up as a as a transparent, firewall, there's other transparent things that we could swap firewall with as well. But I figured a firewall was understandable enough that most people who are non-networkers that they're coming from an enterprise background or similar they understand what a firewall is.

I

They may not understand other forms of packet treatments that are that are there and the transparent portion on this is descriptive of a scenario where it doesn't modify the packets it's uh first, it doesn't modify the pack if it's, if it if it doesn't need to, uh but simultaneously the more important part is that it's is that it's not a routable thing. It's not like you're, not able to address the firewall in the middle of the user plane. Despite the fact, it's still getting applied to the to the user plane itself.

I

So, in short, that's the uh that's. The recommendation is to have something in bad respect.

I

um Other areas where this could be useful as well is when you start looking at like culture use cases, then the in l2 you can put in a bump in the wire arp responder or when you're looking in some security products, you may decide to put something other than firewalls, such as an intrusion, detection system or some form of data lodge prevention.

I

So, in short, that's what uh that's. What this particular thing uh tri tries to represent is that uh bomber, the wire firewall that lives within.

A

Kubernetes cool.

A

uh Any questions comments, thoughts.

C

So I think this one will be interesting and tricky um depending on like how we describe the firewall and what the firewall best practices are. I can tell you like this thing right here is something where non-cnf stuff, but just typical, kubernetes things.

C

If you look at books, use case around being able to provision new nodes, and um you know taint nodes, migrate, pods, all this kind of stuff do upgrades things like blue, green and whatnot become really um tricky when um you have legacy like firewalls as a bump in the wire, um depending on how you pin hole, so um you know like if you don't like he's, got a firewall controller there. So assuming you're doing everything programmatically, then maybe it's part of like a bigger workflow, but um I think this one's use.

C

This use case is interesting because it sounds very, very simple, but it causes heartache depending on how you provision things um and it's like things that are uniquely complicated to kubernetes versus how I would do things in a legacy in a way. So that's my my comment.

C

um I think that we say this is um this is simple but like you'd, be amazed when you try to spin up 15 new nodes in a new subnet, and if your firewall team is not part of the cloud team, what does that look like to get those new prefixes added to the acl like it can be interesting.

I

Yeah, I said simple, but I didn't say easy, so I know this is a particularly uh tricky use case for for kubernetes and if you're, if everything is done through through cni you're, just following cmi policy, then uh you're in good shape.

I

But the moment you try doing something, that's outside of cni policy, then that's that's where these things get really really tricky and the kubernetes form of networking doesn't lend itself well to this form of transparent uh of this transparent communication so like so, I thought this was a pattern that it wasn't just going to be. Oh yeah.

E

I

We support it already hit the check box on ed, where we're good, but there's some there's some there's a definitely a gap here in terms of how do you achieve this within kubernetes, but I think it's also something that we need to work out how to solve properly, because we're going to see this pattern appear over and over and over again, and the risk is when we see other things that are created in cns.

I

Besides bumping the wire firewall, we're going to see this thing reinvented over and over again so establishing a clean set of patterns and working out when they do or do not apply around. This use case, I think, would be uh would be valuable.

G

Yeah we, we should probably throw another use case or two at that service addresses or another one that crop up. So I've got a service. I have multiple up links. The up links have inter individual interface addresses and so on. Basically, it's the same. I've got more than one route on a link, kind of thing that I think you're talking about it yeah. I agree. It's a perfectly sensible thing.

G

I

And this one doesn't cover additional advances. On top of that, like what do I do about load balancing? uh What do I do about uh if I want to chain more than one thing, so there are still other complexities on there, but I figure we start something. That's that's simple and then we can work up the chain of complexity over time, but definitely add it to either the the document.

I

uh I've moved it to a suggest mode, so we can track the the changes.

I

uh But if you add it to the document- or you add it to the discussion- that's linked to- let's link to it, then I'll make sure that gets added to the document and once we're happy with where the document is at, then I'm happy to convert it to markdown and submit it as a as as a part of the repository.

C

Yeah one last piece on this too, um just kind of we'll pull together several points. um One is the concept that a cnf doesn't necessarily need to be a data plane intensive um thing, so I can tell you like. Coming from, like the service provider bias, um we run things like network orchestrators and kubernetes that have really weird and picky egress um requirements right like they need to have reach to certain parts of our network, so um a best practice could just be um egress gateways, or um you know something like that like.

C

I think this is interesting, because I think multiple best practices, which is another point we were discussing earlier, could um spawn from this depending on the type of workload. So this comes back to um the glossary a we need to define what a c and f is. um So that way we can talk about how a best practice actually applies to it and then b. um I don't think a c and f is just what we typically want to like focus on, which is like some.

C

You know data plane, accelerated, really cool high performance network function and just simple things like um I have a network function that has layer 3 requirements, which means that it actually cares about the node ip addresses its source. So then, if I have this transparent firewall in the way, how do I do basic cloud native operations when layer 3 is constantly biting me in the.

I

Butt yeah that that makes sense as well, and um this this also becomes very relevant when you start looking at some of the the 5g use cases. uh When you look at the next generation of uh packet core that that's uh they're trying to to create and as we start to look at some of these things like some of the components around authentication and authorization are decomposed into multiple parts, and some of them are just.

I

This is a database that has state- and this is uh this- is something that will perform some authentication action for you and then the result of those programs, a a data plane somewhere and so over over time. Like that, that thing can look like a bump in the wire firewall to whatever it is.

I

It turns out to be a tunneling protocol or or something similar, but to the, but it still needs to be addressable in a way to the rest of the infrastructure, so that you can still get that control and and manipulation, and this is one of the reasons I try to avoid. The term uh cnf in the discussion was to avoid the argument about what is a cnf versus what isn't.

E

I

Instead, to focus on the heart of the matter of, what's trying to be and what's trying to be solved and one of the things we've done in the network service mesh community is we stayed away from the term, even though we we use the term cnf, but when we start talking about it in a concrete area, we talk about things like network services, something that provides a network service or an endpoint, something that you actually connect to and uh multiple endpoints uh can uh can provide a network service that you can then scale up those endpoints up or down, but they're all part of the same umbrella of network service, whatever you're trying to connect to so so I tried to use something along those lines, as opposed to something that's a little bit more.

I

uh I don't want to use the term ambiguous because it's not really ambiguous, but it's something that there hasn't been a clear definition around as to where do we want to draw the line between an application or or a.

I

A

And victory just commented, uh I suggest we start using the voting feature of the github discussions to protect prioritize items like this yeah. So all about that too.

A

Cool, um so thanks for kicking off that uh conversation frederick um be great to have other people jump into. um I think that's all we have on the agenda for today is there anything anyone else would like to bring up or discuss before we close for today.

E

Well, the only thing that I I was just thinking about is like the numbering of the use cases. um Do you think that could be a good idea to keep numbers? um My concern is like um I don't know.

E

If, eventually we can confuse like using numbers to on on the use, cases can reflect priorities or misconfuse those things um I mean I I, the other thing is like which number you can give to a use case when you have to multiply multiple use cases at the same time like the merge, the merge use the merge number or like the number that you use when you submit it or things like that, but I think eventually we can avoid using numbers, could be we better. I don't know that's.

C

On this point, victor, um we did say in the past that best practices would need to tie to use cases. I think it's safe to think that certain best practices could potentially tie to multiple use cases too, so some type of metadata is um probably warranted. So that way we can have some mapping um and maybe like use cases and best practices, get updated with like a little table at the bottom. That say what they apply to you.

C

I don't know that might be over complicating it, but um the big thing is is like you know, I look at a use case and I'm like okay. How do I solve this? What what is the metadata that leads me to the best practices that would help me overcome the challenge, especially if there's not a singular best practice, but maybe there's multiple ways that the group at large deems is valid. You know means of solving it. We probably want to plan for that.

E

Okay, so so you think that using like a number we can group- or at least I.

C

Don't I don't, I don't know if that's the right way or not like. uh um I just know that metadata so that we can track what goes to what is at some point going to be relevant and the longer we go without it, the harder it will be. If you know we have lots and lots of stuff that we then start saying how do we map all this together?.

A

Yeah, I guess we originally started giving numbers, because it's something a lot of like in kubernetes. The caps are all numbered to um just yeah to make it easier to keep track of and in terms of like merging them, I mean if we're having like hundreds at once, it might be a problem. I think, at the scale we're at right now um we can still kind of keep track of the numbers I mean if kubernetes can keep track of the numbers for caps. I think we could probably come to if we find it's a problem.

A

um Maybe we rethink what we're doing.

C

Yeah, I don't think it'll be an issue with merging bill. I think it's just um you know. Let's say I have four different ways to solve network orchestration like how do I know which best practices I could go and look at to figure out like the one that I think.

E

C

My use, like my flavor of the use case, like I, don't think we're going to have to worry about merging or this and that it's more just you know, how are we going to map use cases to best practices and.

G

They have the file names.

C

G

Names that you.

B

G

B

On that and jeff uh robbie had started on that with the document around best practices, and I I think when it gets to a point where we're saying someone is interested in finding best practices, they can go. Maybe the way the routes you're saying in they could just look through, but hopefully we have hundreds and that may be too much so they may go. Let me go into the document and I want to go down to a category and then you say: oh there's, two options listed or four options to solve this problem.

B

Let me read on all of those and then I'll pick the one that fits me the best.

B

So we can organize it. I think as they come in to place, I don't.

E

Know if this is.

B

This yet yeah yeah yeah, that's fine.

A

So yeah, I guess victor, that's a good point like how do we want to map it kind of to this? So if you have any ideas, um it'd be great to hear them or a better way of doing it.

B

And, and with the the idea with the best practices, every best practice should link back to a use case. So if, if you're going in that way, but you could also do the other way we could, we could always update a use case and reference best practices that have been adopted.

B

So we we can point to those and do a pr against an existing use case. Maybe make a section on those and and point to one or more best practices.

C

Yeah, I don't think we have to solve it right this second, I'm just saying at some point like looks vox pr, like I said, is going to spawn multiple best practices. I think so.

E

C

Sure that there's some type of mapping to all the ones that come around platform, life cycle management, etc um and yeah. Even if it's just retroactively like there's just a list- and you just put the links in- I don't know- or file names to ian's point. It doesn't really matter to me. I'm just saying um assume: you're, not someone who regularly frequencies calls or works and get with us, and you come in because you want to figure out how to do something right.

C

um That should be like our target audience that we're really trying to make this consumable for.

B

I I um I agree, and, and maybe it's going to be multiple target audiences, just someone that's actively doing very specific development may come in and go I'm looking for this one type of thing. They may go right into the best practice list and then someone else may say here: what's the closest use case, so I I think we need. um You know the big specific list, as well as the higher level and maybe even another.

B

One would be if you have a list of use cases that we don't have but are kind of related, because some of the the topics, including what frederick was talking about for the the inline firewall there can be more complex ones. So, if someone's coming in and saying here's the type of thing that I'm having problems with and then find their way down to, what we've written is is the most important.

E

So the other thing that I was thinking is like this seems to be like a change in the templates. Maybe we can I mean if the new use cases are not at least having a one best practice.

E

I don't know if that makes sense to include as part of the best practices right, because.

E

I mean just talking about that. The new the new use cases has to at least has one.

A

A

Sorry, can you say that.

A

A

Victor, are you still there.

E

Yep sorry: what was the question.

A

Oh sorry, um can you just say: wait say that again or did you.

E

Oh no, no, the only thing that I was saying is like. uh Maybe we have to modify the template that we have it and.

B

E

Include like a section where we uh need to provide the best practice which is implemented in the new use case, yeah yeah. I believe we can truly think.

A

Okay, yeah, so maybe that's something to think the future, how we like coordinate them together.

A

Cool, um so we are at the top of the hour. I don't want to go over um just a reminder if you haven't voted yet, please vote by tomorrow at midnight pacific. um Any questions, please feel free to reach out to me. Otherwise, thank you for joining today and have a nice week.

A

Thank you. Thank you. Bye-Bye happy monday.

A

A