[Eastern Time]

[Enrico Fermi Institute] 11:00:03
Cool Dale is indeed here

[Enrico Fermi Institute] 11:00:25
We'll wait a couple more minutes for folks to filter in

[Enrico Fermi Institute] 11:01:41
oh!

[Enrico Fermi Institute] 11:02:29
e

[Enrico Fermi Institute] 11:02:37
Yes, slides

[Enrico Fermi Institute] 11:03:24
okay, we're gonna get here started here just a minute.

[Enrico Fermi Institute] 11:03:28
So thanks everyone for making it to the to the final day of the workshop goal.

[Enrico Fermi Institute] 11:03:35
For this morning will be to do a discussion about number of topics, one which is counting the other would be to continue our pledging discussion.

[Enrico Fermi Institute] 11:03:50
assuming, you know, of all the appropriate people are here for that.

[Enrico Fermi Institute] 11:03:53
another topic. We wanted to cover was security, you know, both on the on the clouds and the Hpcs.

[Enrico Fermi Institute] 11:04:02
And I think ultimately the the stuff that we cover this morning will inform for poly recommendations that we would, that we would make as part of the report in the afternoon.

[Enrico Fermi Institute] 11:04:17
I don't expect us to take the full 2 h may not even take the full 2 h this morning.

[Enrico Fermi Institute] 11:04:26
but in the afternoon we'll talk about the We'll have a presentation from the vera ribbon folks, and then we'll I think we have a couple other minor topics any other, business, and and sort of next steps for work to go with with the us.

[Enrico Fermi Institute] 11:04:47
see So about 20 people. Online So that's probably good to going here.

[Enrico Fermi Institute] 11:04:56
So yeah.

[Enrico Fermi Institute] 11:05:01
Dirk, did you want to make comments about this one?

[Dirk Hufnagel] 11:05:03
yeah, I I can. I can. Get the started. So as as with previous slides, so that the green is is a question that's directly copied from the charge and the the question, was what can the us at less than us, Cms: recommend to the collaboration to improve the utilization of commercial and Sbc

[Dirk Hufnagel] 11:05:23
resources, and both in terms of enabling more work for so we can run there, and also improving, and also improving the cost effectiveness of using these resources, and and one thing that that was all that already came up in the previous days was but we'll have a dedicated slot

[Dirk Hufnagel] 11:05:44
for here, so we'll see how much excellent discussion we're good is that at the moment these resources are nice to have say opportunistically, they give us some free cycles where we can run some things but they're not included.

[Dirk Hufnagel] 11:05:58
In the pledge, so we don't get full credit for them, and we don't them down included in the planning.

[Dirk Hufnagel] 11:06:06
another. The thing is that to make use of sites like the Lcf.

[Enrico Fermi Institute] 11:06:08
Let's

[Dirk Hufnagel] 11:06:12
Specifically, but also other opportunities that are available in cloud.

[Enrico Fermi Institute] 11:06:15
Okay.

[Dirk Hufnagel] 11:06:16
And then our own good sides potentially, we need Gpu payloads.

[Dirk Hufnagel] 11:06:20
We need some way to utilize, deployed Gpu resources on the cloud side We had a lot of discussion yesterday on that in the cost section and doing the networking the the pick, big worry a big part of the where you're on cost on cloud is egress.

[Enrico Fermi Institute] 11:06:24
See.

[Enrico Fermi Institute] 11:06:27
Okay.

[Dirk Hufnagel] 11:06:41
So minimizing egress or negotiating some way, be the peering agreements.

[Dirk Hufnagel] 11:06:49
Either subscription. Where this cost is basically removed or targeting cloud resources that don't have egress charges.

[Dirk Hufnagel] 11:06:57
We basically want to avoid the the egress costs that currently are, they're not dominating.

[Dirk Hufnagel] 11:07:03
But they are large factor of the whole. Cloud cost calculation, and on the Hbc side the focus will be on reducing operational overheads, especially on the Lcf.

[Dirk Hufnagel] 11:07:19
Side, where this is still a little bit of an on D area in terms of figuring out what the what the final operational model will look like, and then another way to reach cost and make make it easier to operate hpc resources is to get sizable storage allocations because it makes

[Dirk Hufnagel] 11:07:39
something simpler.

[Dirk Hufnagel] 11:07:47
We could move on, Maybe at this point, if you're talking about accounting, which is, which is part of like having this these types of resources be a fully equivalent player, I mean, pledging is one side but then you need to take care of of accounting to make sure that you actually deliver what

[Dirk Hufnagel] 11:08:06
you promise to deliver. We have a comment. I guess we can.

[Dirk Hufnagel] 11:08:10
We can do that right right away

[Enrico Fermi Institute] 11:08:14
Yeah, feel free to jump in at any time.

[Dirk Hufnagel] 11:08:16
Eric

[Eric Lancon] 11:08:17
Yes, good morning. Before we change the slide. I think the first recommendation would be that Atlas and Cms both work on the

[Eric Lancon] 11:08:35
Having a Gpu, or whatever or accelerator find the payloads. Because if you don't have a generic software, you you would not be able to use at hi resources

[Eric Lancon] 11:08:56
And if they are very restricted, these resources to a specific kind of application, let's say, just specific.

[Eric Lancon] 11:09:07
evolution, because it's suited to to Hey, Gpus, it There would be no way to get them plagued, because pledges are as they, are for now.

[Enrico Fermi Institute] 11:09:13
Okay.

[Eric Lancon] 11:09:19
It's for all will work close together. It's not greatest by type of workflow, so

[Dirk Hufnagel] 11:09:28
Yes, that that's that's one problem, And we can discuss this later that that the moment the pledges I mean, you can.

[Enrico Fermi Institute] 11:09:33
Sure.

[Dirk Hufnagel] 11:09:34
You can pledge across Cpu architecture. You can see that because you can.

[Dirk Hufnagel] 11:09:39
You can run your normalized, whatever this be. Right now. Hs: 0 6 in the future will be half score.

[Dirk Hufnagel] 11:09:45
You can run your benchmark, you get your Cpu speed it.

[Dirk Hufnagel] 11:09:49
It includes some averaging for the airbus, but at least it's in principle possible.

[Dirk Hufnagel] 11:09:54
Gpu is more tricky, because how do you account for that?

[Dirk Hufnagel] 11:09:56
Do you give like a 20% bonus? Do you look at the most commonly used workflow?

[Dirk Hufnagel] 11:10:02
And I think that's a that's something we can't decide here.

[Dirk Hufnagel] 11:10:06
That's something we'll have to discuss with Wlcg how they once we have these, these, these workloads I can use Gpu and get a benefit from them.

[Dirk Hufnagel] 11:10:16
How does the the how do you factor? That in and the performance normalization?

[Dirk Hufnagel] 11:10:21
Is there gonna be some extra factors? Is a different category that you pledge?

[Dirk Hufnagel] 11:10:25
I I don't know. I don't have the answers, but it's it's an area that that needs to be discussed.

[Dirk Hufnagel] 11:10:30
But Wlcg

[Eric Lancon] 11:10:33
Yes, but the recommendation to the experiments would be first to actually right.

[Enrico Fermi Institute] 11:10:39
The

[Eric Lancon] 11:10:41
The development of a Gpu friendly software

[Dirk Hufnagel] 11:10:45
Yes, I mean, we're doing that, I mean, there are 2 avenues and cms at least at 2 avenues, and and that's that's the machine.

[Dirk Hufnagel] 11:10:53
Learning that we talked about yesterday, which is that that might be the most common at the moment between Cms and Atlas, because similar training frameworks, and so on.

[Dirk Hufnagel] 11:11:04
That you could use in the framework itself. For obvious reasons.

[Dirk Hufnagel] 11:11:08
This is somewhat distinct from each other. But we're both looking at that.

[Dirk Hufnagel] 11:11:12
Cms: maybe is a little bit more advanced than Atlas because of the the the push from the hlt, and the deployment of Gpu Also resources there.

[Dirk Hufnagel] 11:11:21
But I'm sure that Atlas is also at least looking at this

[Enrico Fermi Institute] 11:11:31
Okay.

[Dirk Hufnagel] 11:11:36
Lincoln. Can you go to the next slide?

[Dirk Hufnagel] 11:11:39
If that comment was addressed. So, looking at accounting, so maybe at this point, this is a good time.

[Dirk Hufnagel] 11:11:46
We have an invited contribution for benchmarking, because for accounting and for pledging one of the prerequisites is that you have to know what you're actually pledging in terms not just not in terms of course, but in terms, of some normalized numbers, hs, 6 or in the

[Dirk Hufnagel] 11:12:05
future have score, so that there that Maria offered contributed Talk about benchmarking of Hbc I think

[Enrico Fermi Institute] 11:12:06
Good.

[Dirk Hufnagel] 11:12:16
And David wanted to give the toocacy connected

[David Southwick] 11:12:20
yeah, I am. Can you hear me?

[Dirk Hufnagel] 11:12:21
Okay, Yeah, Do you want to? Okay, Do you want to share the slides or otherwise Link can also share

[Enrico Fermi Institute] 11:12:22
Yeah, we'll I'm not sharing anything

[David Southwick] 11:12:25
Okay. Just like my headphones have

[David Southwick] 11:12:35
Just a second here

[David Southwick] 11:12:38
Okay, you can still hear me. Great. So I do have just a few short slides.

[Enrico Fermi Institute] 11:12:41
We can.

[Enrico Fermi Institute] 11:12:45
Okay, okay.

[David Southwick] 11:12:45
That I'd be happy to hear

[David Southwick] 11:12:49
See if I can do this

[Enrico Fermi Institute] 11:12:51
Sure.

[David Southwick] 11:12:59
Yes, hmm.

[David Southwick] 11:13:06
Okay.

[David Southwick] 11:13:13
Okay, Do you see the? So I do. Okay? Great: So Yeah, I've got a few comments, and just to share a bit by what we're doing.

[Enrico Fermi Institute] 11:13:15
We do. Yep. Looks good

[David Southwick] 11:13:27
With Hp. See with the head benchmarking Last couple of years I've been collaborating with the hex benchmarking working group really to take the this replacement model, for he speculated 6.

[David Southwick] 11:13:49
And really developed this that so that it can work on Hpc.

[David Southwick] 11:13:52
As well, cause I'm I'm sure, sure many people are very familiar with this, and how it was done in the past.

[David Southwick] 11:13:57
It was meant to be as as similar as possible to the bare metal works.

[David Southwick] 11:14:04
Notes that Wcg. Was using. So it was Vms, or at some point nested containers.

[David Southwick] 11:14:11
And things like this that are no way compatible with Hbc: So, but in a bunch of work to to make this as lightweight and user friendly as possible, So it's totally rootless.

[Enrico Fermi Institute] 11:14:23
Okay.

[David Southwick] 11:14:27
Now, we're using switch to singularity images, and then also a bunch of quality of life things that that allow you to use it on sites that have, you know, don't have wide area in networking tool Okay, things like that So it's really, been a big big it for

[David Southwick] 11:14:46
the last year or so a bit about the sweet self.

[David Southwick] 11:14:51
I'm sure some of you are No, you're over.

[David Southwick] 11:14:52
Maybe I've run this already since we've been distributing say, proof of concept or release candidates here for the last couple of months.

[David Southwick] 11:15:04
basically it's well, I love this. I already shared, But it's now sort of flexible thing that, and you can run on any hardware.

[David Southwick] 11:15:14
You can see, I've got a small graphic on the right right here, or the suite itself is an orchestrator that will go and collect metadata of whatever hardware it's running on and control the array of benchmarks you want to use so on that the bottom

[David Southwick] 11:15:30
part is graphics. There's a couple different benchmarks had spec of 6 is one of them He score, which is the the the candidate replacement for it or I don't know if I can call it Canada anymore.

[David Southwick] 11:15:46
but you can easily plug in other mitch marks as well.

[David Southwick] 11:15:49
So this is the tool we've been using on Hbc: So a bit about that, This effort, like I said, started.

[David Southwick] 11:15:59
I guess more than a year ago. So I and the initial presentation of the Hpc.

[David Southwick] 11:16:07
Work that was during chat 21, and at that time we had just done large-scale deployments, so doing several 100,000 core campaigns, and at that time we were looking at comparing New Amd.

[David Southwick] 11:16:22
Mdc. Views that were available widely on Hpc.

[David Southwick] 11:16:26
Sites, but not yet widely accessible, and from so we did a comparison of that, and stability, studies and whatnot.

[David Southwick] 11:16:37
So that was interesting. The first step. What's happened since then is we've had a lot of software become available from the experiments.

[David Southwick] 11:16:48
obviously the first look at the run. 3 workloads, but along with that there's been a bunch of heterogeneous development from basically all of the experiments.

[David Southwick] 11:16:57
So and I mean heterogeneous, both in compile compile codes and in accelerators.

[David Southwick] 11:17:06
So we've got several workloads that are in development for power, and of course, Gpus So we've been using Hbc Then to take these these workloads, or let's say, snapshots.

[David Southwick] 11:17:23
Of them. They're sufficiently stable. I will containerize them in singularity, and then run them at scale on Hpc.

[David Southwick] 11:17:32
And that enables a lot of different interesting studies. So Gpu versus cpu, and then combined for some, we're close there, support that as well as more exotic combination.

[David Southwick] 11:17:46
So arm, plus gpu said Power plus gpu, and things like this.

[David Southwick] 11:17:54
And I know this was discussed. Excellent! Some point yesterday.

[David Southwick] 11:17:56
I think I was listening, in but in case these are now available via the benchmarking suites, and you can run it just as you would on a bare metal machine on Hbc.

[David Southwick] 11:18:12
I think you it was presentation yesterday from Eric about the in Lps: Yeah, I workload.

[David Southwick] 11:18:20
That's also containerized, and can be run at scale on each Pc.

[David Southwick] 11:18:25
however, the configuration they have at the moment is a the the snapshot of that is just single note.

[David Southwick] 11:18:31
At the moment. So if you do want to use that for Npi related, scaling, then you need to run just the workload container and not the the suite, because it well it's not the target at the moment of of Wcg to do Npi and as I mentioned I guess we've

[David Southwick] 11:18:54
got a lot of other quality of life things. So if you have local storage like Cdm Fs, you could take advantage of this instead of room, pull copies and whatnot.

[David Southwick] 11:19:03
But really, as has been set here many times there's a lot of interest in Gpus, and then you can see I got a small slice here in the number.

[David Southwick] 11:19:15
Of course you get from current and next generation Gpus.

[David Southwick] 11:19:21
So there's I love to going in that direction, and we know we're expecting close.

[David Southwick] 11:19:30
so, that being said, there isn't industry, standard Gpu benchmark yet, and at least from the benchmarking side of things.

[David Southwick] 11:19:43
We are kind of approaching it in the same way that we have Cpus.

[David Southwick] 11:19:48
Oh, you use workload set production workloads, or what will be production workloads, And we can generate the score in the same way that we do for head score which is some function of throughput or events.

[David Southwick] 11:19:59
Per second. So this is what we've been using so far with trying to understand the capabilities of a machine that is going to be running Gpu.

[David Southwick] 11:20:13
Only Cpu, and I mean there was a he score work last week, I think several people are probably part of, but a lot of discussion happened there as well of how to account for for these sorts of resources.

[David Southwick] 11:20:34
so how just concluded, saying that we we've been active on Hbc.

[David Southwick] 11:20:38
Benchmarking now for couple years. Use the suite because it's automated running and reporting of large scale.

[Enrico Fermi Institute] 11:20:39
okay.

[David Southwick] 11:20:48
You can do whole part or several partition benchmarks.

[David Southwick] 11:20:52
This includes exotic workloads both for machine learning.

[David Southwick] 11:20:56
Yeah, I as well as architectures as well as starting to look at.

[David Southwick] 11:21:03
Let's see, sort of the other services that you get on Hpc: I know it was mentioned yesterday that it issues with scaling.

[David Southwick] 11:21:14
Io bomb workloads. And how can you tell what what's good on a shared filer system that maybe you don't have any information about.

[David Southwick] 11:21:24
So we are starting to develop. And we've got a prototype.

[David Southwick] 11:21:27
Let's see Mark. I say benchmark kind of in quotes, because it's not benchmarking a a compute, unit, but testing the shared filer service.

[David Southwick] 11:21:37
And then from there giving you some feedback on both your workload and let's say, how many nodes you could scale that up to before where it starts locking up that the file system in some way.

[David Southwick] 11:21:53
So that's what's new with us, and a little bit of a peek into.

[David Southwick] 11:21:57
We're doing in where we're going and I'm happy to answer any questions

[Enrico Fermi Institute] 11:22:06
Okay, David. Thank you very much. So we have a couple of hands raised.

[Enrico Fermi Institute] 11:22:10
follow up! Good

[Paolo Calafiura (he)] 11:22:12
good morning, everyone. So the the first you said that you said that the there are no industry standard Ml.

[Paolo Calafiura (he)] 11:22:23
Benchmarks, and I think that is still accurate.

[Paolo Calafiura (he)] 11:22:26
But I want to meet. I want to be sure you guys are aware of Ml.

[Paolo Calafiura (he)] 11:22:30
Per which is becoming a little system, and it's not so.

[Paolo Calafiura (he)] 11:22:36
So okay. So that's that was the first. The first comment.

[David Southwick] 11:22:36
yeah.

[Paolo Calafiura (he)] 11:22:41
The second comment is that the and I mean, I I You guys have a very difficult job, because what we are seeing in Tc.

[Paolo Calafiura (he)] 11:22:52
Is that the same software? I mean the same out of it around on different software, on different platform, performs quite differently.

[Paolo Calafiura (he)] 11:23:04
So, if you have, a if you have a fast parameter or simulation, we should run that with alpaca, or with the kuda, or with the of well, cool, that is a different problem without Packer. With caucus, you'd get different different performance, of the same code in

[Enrico Fermi Institute] 11:23:13
Hmm.

[Paolo Calafiura (he)] 11:23:22
principle on day for different machines depending. What what portability layers to probability to you.

[Paolo Calafiura (he)] 11:23:31
So I wanted to ask you if you have a settled on a platform, for like parallelization platforms, but to use, or if you are taking the world, you know you're taking a mix so what what's your problem

[David Southwick] 11:23:45
so I don't think it's settled at the moment, and of course this is a a popular question of what sort of optimization targets you're using for your workloads.

[David Southwick] 11:23:56
And I mean not just for

[David Southwick] 11:24:01
Translating the Cross architecture. But even within the same families of of of units.

[David Southwick] 11:24:09
So at the moment, I I think the method we have is take the minimum compatibility, so that we don't having, you know, 1020 different versions of of the workplace Okay, want to have the same thing that we can run everywhere and and sort of take all of these variables out of the

[Enrico Fermi Institute] 11:24:18
Okay.

[David Southwick] 11:24:30
equation, But that being said, I don't think there's a clear answer on on the proper way to do that yet, and it's not not so.

[David Southwick] 11:24:42
Let's say.

[Enrico Fermi Institute] 11:24:48
Okay, hum. See, Steve has this hand raised

[Steven Timm] 11:24:51
Yeah, I was just wondering if the toolkit is available anywhere for download that have been

[David Southwick] 11:24:58
Oh, good, Yeah, Absolutely So the let's say I'll just go back.

[David Southwick] 11:25:04
I think I have a link to. And so the benchmarking suite itself, that is it. While all that is open source.

[David Southwick] 11:25:13
It's on git lab at certain. So flash, hip, benchmarks, and then the suite is the the project down there.

[Steven Timm] 11:25:15
oh, okay, I see it. Okay.

[David Southwick] 11:25:22
I have the link on the screen, the yeah, the other benchmark I was talking about, for let's say services.

[David Southwick] 11:25:30
This Io Benchmark. I don't have the link on there, but I can share this afterwards.

[David Southwick] 11:25:36
It's in a That's right. Prototype state right now.

[David Southwick] 11:25:40
We have been working on this year. It doesn't cover all the things you can throw at it yet.

[David Southwick] 11:25:47
But yeah, you can download it and play around with it.

[David Southwick] 11:25:51
And I should mention that the idea for this will really is from, I I think I don't know if he's in in the room now, but it I've seen, and the previous days so shut up. To.

[David Southwick] 11:26:04
Him, I guess.

[Enrico Fermi Institute] 11:26:07
Okay, Okay, go ahead.

[Steven Timm] 11:26:10
Hello! There was no different so they have Benchmark 3.

[Steven Timm] 11:26:14
That's no different than the when they were running on regular note.

[Steven Timm] 11:26:17
Then

[David Southwick] 11:26:18
Yeah, Yep: exactly.

[Steven Timm] 11:26:20
Okay? Good. And then there may be a newly special one.

[David Southwick] 11:26:25
Well, there, yeah, like I said, the there was a workshop last week on this discussing.

[David Southwick] 11:26:33
you know how to choose the final versions and the waiting and whatnot.

[David Southwick] 11:26:37
So there's a little bit more qualified people around, I think, to answer specific questions on on that.

[David Southwick] 11:26:43
But it's in progress. Yeah.

[Enrico Fermi Institute] 11:26:47
Okay.

[David Southwick] 11:26:48
There will be another version, I guess, with the the let's say, gold standard for okay for the Benchmark suite.

[David Southwick] 11:26:55
That's decided

[Enrico Fermi Institute] 11:27:02
Sure.

[Dirk Hufnagel] 11:27:04
yeah, I just had it quick. Question. You said, You're you're benchmarking cpu plus gpu and also, Gpu workloads.

[Dirk Hufnagel] 11:27:13
Now I can see. I mean hapscore, and have specs zeros on cpu that's like well established.

[Dirk Hufnagel] 11:27:20
You take a mix of experiment, specific workloads and average something, throws something together and get get some average.

[David Southwick] 11:27:26
Yep.

[Dirk Hufnagel] 11:27:26
I have. What do you do for the Gpu stuff?

[Dirk Hufnagel] 11:27:30
Because it's so early, and and the experiments I'll algorithm.

[Dirk Hufnagel] 11:27:34
I know Cms has something, but it's not complete.

[Dirk Hufnagel] 11:27:36
It's not a complete picture. Do you run synthetic stuff, or do you run the very early stuff?

[Dirk Hufnagel] 11:27:42
Because that's the only thing you can do

[David Southwick] 11:27:43
We this there. Yeah, So we're running very early stuff from Cms: There's an Lpf.

[David Southwick] 11:27:50
Which is a bit of a an exact bird, which Yeah, there was a talk yesterday on that.

[David Southwick] 11:27:56
we're we also are using hlt and then as well, the sort of rolling builds from check windows, from mangraph

[Dirk Hufnagel] 11:28:11
Okay, thanks.

[David Southwick] 11:28:13
I guess there's also some other exotic Gpu workloads, but these are from Beams Department.

[David Southwick] 11:28:21
So there's this simple track, or

[David Southwick] 11:28:26
I know ped track is in there as well, but I don't think we have a container for Patrack

[Dirk Hufnagel] 11:28:31
so it's it's early going. So the numbers you get might not necessarily be representative of ever we end up running in production later

[David Southwick] 11:28:40
Exactly so. I mean, there's a lot of results results already, since we you know, we use the the suite as a reporting tool as well.

[David Southwick] 11:28:48
So it's it all gets pushed up over Amq into Cabana, and then you can All the workloads are hashed so you can compared performance across every node that it's run on with the same version of the world Okay, so at least with.

[David Southwick] 11:29:04
You know that published version? If you have your own bill, let's say you can track You can compare device to device like this, but you're right.

[David Southwick] 11:29:13
I mean, these are really, or let's say, snapshot releases of some of these.

[David Southwick] 11:29:19
So they will change whenever it's decided that that's going to be a production.

[David Southwick] 11:29:25
Or let's say, a a final version validated in some way.

[David Southwick] 11:29:28
Yeah.

[Paolo Calafiura (he)] 11:29:29
so starting to jump in in Cc. We have assembled the kind of what we think is A is a cross-section of representative applications, and of course we have no, we have No.

[Paolo Calafiura (he)] 11:29:42
Standing is just like asking around. I wonder if we should compare now, and we can see if if we, if we pick the same, and under which configuration maybe we should have an offline discussion between our groups

[David Southwick] 11:29:54
Sure you're using workforce or like off the shell benchmarks

[Paolo Calafiura (he)] 11:30:02
No, no, we're using using a Hp workload.

[David Southwick] 11:30:06
Yeah.

[Paolo Calafiura (he)] 11:30:08
So simulation tracking. We're not doing machine learning workloads yet.

[Paolo Calafiura (he)] 11:30:11
So that's that's something that's something which is not.

[Enrico Fermi Institute] 11:30:13
Okay.

[Paolo Calafiura (he)] 11:30:14
But we should we should, we should compare notes also because of this dimension.

[Paolo Calafiura (he)] 11:30:18
This is not only the workload it's the is, the is the software software platform you use, which makes 1 one configuration different from enough.

[Paolo Calafiura (he)] 11:30:28
Anyway. Bye. Shut up

[David Southwick] 11:30:30
Yeah, I know we can click offline

[Enrico Fermi Institute] 11:30:35
A quick question, How long do these benchmarks take to run

[David Southwick] 11:30:38
So if there is some of the the Gp ones can be fast in the mat in the order of I don't know 20 to 60 min from the some of the Cpu ones are much longer depending on what the experiment code owners have put forward.

[David Southwick] 11:30:58
As well what they you, as a representative set. So I think the default block or Cbu only, and the current release candidate is something like forward to 6 h.

[David Southwick] 11:31:13
But that's many workloads they do back to back, and you run 3 3 iterations of each to get an average and get rid of outliers I'm not sure what that will look like or Gpu because all all the work was that talked about for Hpc that are

[David Southwick] 11:31:32
not sort of run 3 standard ones. These are optional things that you can elect to run on with the suites They're not included by deal.

[David Southwick] 11:31:45
They are available. You just have to use a little bit different configuration, which is included in the

[Enrico Fermi Institute] 11:31:53
Yeah, I guess the thing I'm wondering. And maybe this is just a broader general question for everybody here, I mean, is this the kind of thing that we want to start incorporating?

[Enrico Fermi Institute] 11:32:02
Integration process for Hpcs right? We go to, you know.

[Enrico Fermi Institute] 11:32:07
Stand up pro mudder, and then the next machine should make sure.

[Enrico Fermi Institute] 11:32:11
We run these benchmarks as part of that integration.

[Enrico Fermi Institute] 11:32:13
So we start getting, you know, the benchmark numbers in place, and then maybe that helps eventually with the pledging and that sort of thing

[Dirk Hufnagel] 11:32:23
I I can just say what we're doing right now, because even though they are optimistic, we still at the end of the year cannot compile the usage and and have spec with 6 to come just to have a comparison and to see what the what the big picture looks like and we just went through

[Dirk Hufnagel] 11:32:40
that exercise in 21 for all the Hpc.

[Dirk Hufnagel] 11:32:44
That we're using in the Us. And basically what I'm doing right now.

[Dirk Hufnagel] 11:32:47
I look at the Cpu, and compared to what?

[Dirk Hufnagel] 11:32:49
What others have, benchmark on the on, on, on, and usually you'll find a number where you can come up with a defensible.

[Enrico Fermi Institute] 11:32:56
The

[Dirk Hufnagel] 11:32:58
Have specs 6, but I mean, especially if you want to get to A If we have really pledging the resources with that, and it becomes relevant.

[Dirk Hufnagel] 11:33:07
I think we we need to run the benchmarks.

[Dirk Hufnagel] 11:33:09
Maybe maybe not right now. But but but if once we get to that point, I think we we should to get a better number.

[Enrico Fermi Institute] 11:33:17
Yeah I mean, it seems like we ought to have a plan for it. I mean, even if even if the content of the benchmarks themselves change over time just to have that kind of in our minds, and kind of in the pipeline for one we integrate these resources.

[Dirk Hufnagel] 11:33:17
Okay.

[Dirk Hufnagel] 11:33:30
We have a couple of race since Linkedin. Maybe we should get these comments

[Enrico Fermi Institute] 11:33:34
Yeah, Okay, Andrew set his hand up

[Andrew Melo] 11:33:39
yeah, I was just gonna point out, I mean, so so Dirk talked a little bit earlier about you know, like, How do you?

[Andrew Melo] 11:33:45
How do you you know account for the Gpus and in the Hep score?

[Andrew Melo] 11:33:52
And you know, he suggested, maybe you give like a 20% bonus, or something like that.

[Andrew Melo] 11:33:55
I think that I think that what makes sense, and I argued this at the Heps for meeting last week is that you can't really benchmark machines with just one single scalar anymore.

[Enrico Fermi Institute] 11:34:00
Cool.

[Andrew Melo] 11:34:04
Right? So I think you're just gonna have to be some sort of tuple, you know, per machine to have these different accelerators on it.

[Andrew Melo] 11:34:11
And I was gonna also point out that that while you know we're we're working on, I guess you can call it hip score 22 with with run 3 work close the the number that pops out of he score right now is it waited

[Andrew Melo] 11:34:26
after. It's just the feature. So, at least initially, what will be pledged You know the this head score unit that will be pledged will will only be taken into account the the Cpu which one.

[David Southwick] 11:34:29
Okay.

[David Southwick] 11:34:37
Thanks, Andrew, and I'd like to add on to that, since we do have this reporting automated reporting, and it gives you the the Json.

[David Southwick] 11:34:47
Okay, all of the work ones. Yes, it will give you a single say, Hepscore value, but it also gives you the value.

[Enrico Fermi Institute] 11:34:49
Yes.

[David Southwick] 11:34:55
Every workload, So if you once like, you're just interested in htt, or whatever it is, you can get the number on that were on that machine for that benchmark, and you can go compare just then the benchmarks you're interested in this is also already available.

[David Southwick] 11:35:14
So it is a bit, I guess, in that way

[David Southwick] 11:35:20
A bit more fine grain than than what we had in the past.

[Enrico Fermi Institute] 11:35:25
Okay, question or comment from Ian

[Ian Fisk] 11:35:29
yeah, it was only that I It was second, that I think it's valuable to be using the benchmarks as we begin to commission the multiple Hpc Sites also I think, in addition to having the the benchmark that tells you a number about how well they're

[Ian Fisk] 11:35:43
performing. I'm wondering, if it also starts a purpose as sort of what we used to think of as the site availability tests for some things There's a diversity in the workflows and that if they all succeed and give reasonable numbers.

[Ian Fisk] 11:35:54
You also have a reasonable expectation that the site is pretty well configure.

[Ian Fisk] 11:35:58
Again.

[Andrew Melo] 11:36:03
so so So a fun anecdote about that, email, you know, we we actually did see, some of this where someone was bench benchmarking Dean and saw the you know this machine.

[Andrew Melo] 11:36:15
That you know they They knew what the head score should be, for that machine was about half or you know, 75% of what they were seeing, and it turns out that the cooling of that racket failed and the machine, was actually power throttling.

[Andrew Melo] 11:36:24
So it was something that yeah, people are able to say, Hey, this machine isn't working right just from looking at these numbers

[Ian Fisk] 11:36:30
alright.

[David Southwick] 11:36:33
Yep.

[Ian Fisk] 11:36:33
Yeah, I think the the other thing that the as we in the Commission, probably more applicable to Hpc.

[Ian Fisk] 11:36:40
Than cloud is that these machines are much more complicated than we're.

[Ian Fisk] 11:36:44
They're not as sort of simple as a pile of essentially x 86 servers.

[Ian Fisk] 11:36:48
They tend to have more complex services, whether cooling or interconnect, or whatever; And so I'm a more detailed set of benchmarking meetings

[Enrico Fermi Institute] 11:36:59
Okay.

[Enrico Fermi Institute] 11:37:02
So you You have in one of your sites. You mentioned that you know you do the uploading of all the results, and you also mentioned that you have some kind of batch uploader for portal.

[Enrico Fermi Institute] 11:37:13
Secure, workers. Does. That mean that this would work on Lcfs where the workers don't have any, you know.

[Enrico Fermi Institute] 11:37:18
Outbound connectivity would just batch it and upload it from the login nodes.

[Enrico Fermi Institute] 11:37:21
Is that is that the idea

[David Southwick] 11:37:23
Exactly. You know. Sites like this are that's common, but they're not uncommon, either.

[David Southwick] 11:37:30
There's several that we've been working with here in Europe that have a similar configuration, and normally the default case for this is just running a you know the base case is to run it on single node for vendors.

[David Southwick] 11:37:43
Or whatever it is, and when the runs are finished it'll compile the report, and then send it over and queue.

[David Southwick] 11:37:50
But if you don't have connectivity on the machine that you benchmark, then you can collect these.

[David Southwick] 11:37:59
these Jason afterward and do a batch reporting basically Yeah, from a from a gateway note

[Enrico Fermi Institute] 11:38:06
Okay, great.

[Enrico Fermi Institute] 11:38:10
Other questions for David

[Enrico Fermi Institute] 11:38:18
Okay, Thank you, David.

[David Southwick] 11:38:19
Yep thanks.

[Enrico Fermi Institute] 11:38:25
The slides again

[Dirk Hufnagel] 11:38:32
yeah, so.

[Dirk Hufnagel] 11:38:36
So when you, when when we look at that process, so assume we we get some benchmark.

[Dirk Hufnagel] 11:38:40
Now some defensible numbers, and we figure out how we gonna deal with the Cpu problem and work out how we pledge this with W.

[Dirk Hufnagel] 11:38:48
Sog At that point accounting become goes from Nice to have to.

[Dirk Hufnagel] 11:38:55
Actually we have to justify what we're using, and to show that we're actually fulfilling the patch.

[Dirk Hufnagel] 11:39:01
And at that point accounting becomes from from from it comes mandatory. Right now.

[Dirk Hufnagel] 11:39:07
It's optional, because we want to know what we're using.

[Dirk Hufnagel] 11:39:10
But when when the numbers start to matter, when we actually have a pledge in place, then we need to show that we actually deliver that pledge and the current situation is is that Cms, is, is made a push last year or made an effort last year, to to get all the accounting data

[Dirk Hufnagel] 11:39:29
pushed to apple. We still have some problems there with some sites where we're using like multi note jobs where the the system isn't quite aware.

[Dirk Hufnagel] 11:39:38
That there's actually multiple notes behind it. It thinks it's one.

[Dirk Hufnagel] 11:39:41
But but that's some technical, difficulties we're working on and in principle things are are connected Atlas doesn't currently, do.

[Dirk Hufnagel] 11:39:51
This cloud usage is is an open question. What, Fernando? You're not currently pushing the your cloud users data to to apple right or to any great accounting portal

[Fernando Harald Barreiro Megino] 11:40:03
no, I'm not this. Okay, apple that is written in the slides.

[Fernando Harald Barreiro Megino] 11:40:11
That's a solution that, for example, Ryan Taylor from University to Victoria, implemented because he's using a similar model as so using the cloud his private cloud and there he did this K.

[Fernando Harald Barreiro Megino] 11:40:25
Upper, because he needed to push the resources. So there is some solution.

[Fernando Harald Barreiro Megino] 11:40:31
But I don't have experience with that. And I'm not using it at the moment.

[Fernando Harald Barreiro Megino] 11:40:35
And it's also not applicable to. For example, Hpcs

[Dirk Hufnagel] 11:40:41
Okay, and and and on top of accounting, there's also monitoring like operational monitoring.

[Dirk Hufnagel] 11:40:49
We were doing this already, but depending on what integration method you pick for an Hpc.

[Dirk Hufnagel] 11:40:56
this can be tricky. For instance, in in the Us.

[Enrico Fermi Institute] 11:40:56
Okay.

[Dirk Hufnagel] 11:41:01
We we basically overlay a logical side on top of each Hbc facility, basically internally in Cms is this thing monitoring infrastructure.

[Dirk Hufnagel] 11:41:16
And the assumptions that it's it's build on.

[Dirk Hufnagel] 11:41:19
But, for instance, for the the Italian case, with the with the site extension to an to the Marconi, 100 Hpc that they've they chose a different model It's it's it's a site extension so basically, everything, is is that is under the

[Dirk Hufnagel] 11:41:36
T; one side umbrella, and then an accounting can be a bit tricky because you cannot use the the the site name as a as a dividing line between what resources are tier one What resources?

[Dirk Hufnagel] 11:41:48
Are on the Hpc. And then then you kind of have to look at sub so identify us that that basically divide this further into subsides and not all monitoring systems are basically geared to support that when we've done some work on that it's it's not Yeah, But

[Dirk Hufnagel] 11:42:10
this is the problem. And in in cloud, so far, at least, Atlas is, is, is doing their own separate side, and ponder.

[Dirk Hufnagel] 11:42:21
So this this problem doesn't come up. Also, the Cms scale tests basically overlay a separate side on the cloud resources.

[Dirk Hufnagel] 11:42:29
but you could also, imagine if if you do, seamless extension of like, if it's 2 decides that they want to support extension of the batch resources in the cloud, that issue will also come.

[Dirk Hufnagel] 11:42:40
Up. I mean, if if they want to do separate accounting, that is

[Enrico Fermi Institute] 11:42:46
Okay.

[Dirk Hufnagel] 11:42:47
We don't have anyone from Osg here, so I don't think we can get any comment on that

[Enrico Fermi Institute] 11:42:53
And

[Enrico Fermi Institute] 11:43:00
Other comments about accounting. I'm sure we move on

[Dirk Hufnagel] 11:43:14
Okay, Now, we actually have something about pledging, And we already talked about that, and in in the last 2 days, and I don't want to rehash that this discussion here we we talked about the difference between Ac and Dc.

[Enrico Fermi Institute] 11:43:18
Yes.

[Dirk Hufnagel] 11:43:31
And planning, like capacity, integrated capacity, and instantaneous capacity, and and the problems related to that and and how it, how it, how the scheduling of of Hbc and cloud impacts that yeah, also one thing to look is that they're not Hbc: and cloud resources

[Dirk Hufnagel] 11:43:54
they're not official osg. And Egi side, so we don't.

[Dirk Hufnagel] 11:43:58
We don't want to be Don't cut, gigs, tickets We talked, we talked about the G.

[Dirk Hufnagel] 11:44:02
You guys tickets already. And when we talked about the cost to support this resources, you need to set up some unit that supports them.

[Enrico Fermi Institute] 11:44:06
Okay.

[Dirk Hufnagel] 11:44:10
for instance, Cms has that team at formula. So anytime there's problem at the Us.

[Dirk Hufnagel] 11:44:16
Hbc. Sites. We get a ticket for that, and and the question here is, where do we see this going?

[Dirk Hufnagel] 11:44:22
And like, maybe not next year. Maybe maybe not even 2 years.

[Dirk Hufnagel] 11:44:27
But where where we are, where do we want to be in 5 years, like?

[Dirk Hufnagel] 11:44:32
When we asked, Let's say, just before Hrc.

[Dirk Hufnagel] 11:44:35
Starts, up, What's what's the goal here? And then we can.

[Dirk Hufnagel] 11:44:40
This requires discussion with Wsg

[Douglas Benjamin] 11:44:48
in both cases we don't own clouds or the Hpcs right

[Dirk Hufnagel] 11:44:54
Yeah, we don't, we? We don't own them, but we we basically what?

[Douglas Benjamin] 11:44:59
Therefore we are customers of them.

[Dirk Hufnagel] 11:45:02
What are we leasing them? I mean, in some sense it doesn't matter who actually owns the hardware.

[Enrico Fermi Institute] 11:45:05
Okay.

[Dirk Hufnagel] 11:45:08
It matters that you get guaranteed access in some way

[Douglas Benjamin] 11:45:13
Not necessarily.

[Douglas Benjamin] 11:45:17
Right. We are a customer

[Dirk Hufnagel] 11:45:20
Via customers Correct

[Douglas Benjamin] 11:45:22
So we have to deal with the interface layer. That is our community ready. I. E.

[Dirk Hufnagel] 11:45:27
But that's that that's support that's that goes into support that we, since we are customer, we we have to be the middleman for for supporting the the resources.

[Douglas Benjamin] 11:45:27
J. Goes.

[Douglas Benjamin] 11:45:38
And then the pledging right, but the pledging then comes from money that we get to provide the compute.

[Dirk Hufnagel] 11:45:38
So we we are the the interface to the to the experiment

[Douglas Benjamin] 11:45:47
If we have to provide to Atlas

[Dirk Hufnagel] 11:45:49
Yeah, the fundamental difference is that the, the, the entity that owns the resources is doesn't have a relationship with W.

[Dirk Hufnagel] 11:45:59
They basically they have only a relationship with us. And then we have a relationship with Wlg.

[Dirk Hufnagel] 11:45:59
they.

[Douglas Benjamin] 11:46:09
And do you expect that to change? And 5 years? I don't

[Douglas Benjamin] 11:46:15
He served it. Both Clouds and Hbc. Serve different.

[Douglas Benjamin] 11:46:18
The sites serve different masters different. They have different, You know the Hpcs in the Us.

[Douglas Benjamin] 11:46:26
Are responsible to Msf. And do we

[Douglas Benjamin] 11:46:33
Right. Am I missing something?

[Steven Timm] 11:46:34
Okay? Awesome.

[Dirk Hufnagel] 11:46:35
No But But why do we need to care? Because if we get an allocation for a 1 million like a 100 million hours, that's something we can use.

[Dirk Hufnagel] 11:46:45
We'd have no guarantees when we can use it.

[Dirk Hufnagel] 11:46:48
But well, we get something that we'll have this over a period of time.

[Steven Timm] 11:46:50
And

[Steven Timm] 11:46:56
And at least in the Us. And we know for the funding This is the way we're going.

[Steven Timm] 11:47:00
They're not going to be funding the level of blab on computers that they have great.

[Steven Timm] 11:47:06
So we're ready to evolve beyond resources that they own and work on a regular basis.

[Steven Timm] 11:47:13
And the ones that they don't want. They don't own, I mean smaller experiments have been doing this forever, just running up basically everywhere.

[Steven Timm] 11:47:24
Zoom account

[Dirk Hufnagel] 11:47:27
Oh, hello! You had to comment on this discussion

[Paolo Calafiura (he)] 11:47:29
the yes, just to say that I don't think it's particularly productive to discuss what what the you know, one support and what the the the agencies, what the agencies long.

[Paolo Calafiura (he)] 11:47:46
Term plan we need to be ready for what they are telling us now, which is that they want us to use the secondary services, But that that's that wasn't the reason I raised my end I wanted I wanted to bring up.

[Paolo Calafiura (he)] 11:48:00
Another angle, and I don't know if this is just the natural single, or or if you guys, have a senior concept.

[Paolo Calafiura (he)] 11:48:05
So. And this is has to do with this capacity. It's versus power that we have been There's that the or what versus power that we've been discussing.

[Paolo Calafiura (he)] 11:48:15
So in Atlas we have the concept of pledge, and beyond pledge resources, and I I I don't even wanna try to tell you why we have this distinction but to, the the it is historical.

[Paolo Calafiura (he)] 11:48:31
But the the reality is that the pledge in our class is sufficient to process data more than sufficient to process day time to produce a meeting of simulation that would probably be sufficient to process data 2 to one analyze data, but we do rely on a very substantial amount of beyond

[Paolo Calafiura (he)] 11:48:54
pledge resources, which is a which is taking it taken into account, is measured, and the end is not technically pledged, so there is no distinction in my mind.

[Paolo Calafiura (he)] 11:49:07
Between a tier, 2 delivering twice as many resources as they as they are supposed to, and the an Hpc.

[Paolo Calafiura (he)] 11:49:16
Delivering this delivering the same resources. So the question, the question is, is this concept of pledge absolutely fundamental?

[Paolo Calafiura (he)] 11:49:27
Or is it something which is, you know? We start together, we deal with it, and then and then we treat the Hpcs and the you know any resource which which can deliver resources on not on a constant basis.

[Paolo Calafiura (he)] 11:49:42
But on a on a opportunistic way just this, beyond pledge

[Dirk Hufnagel] 11:49:50
I mean, we're doing this now, That's how we're treating Hpc: Now the the question is going going forward.

[Dirk Hufnagel] 11:49:58
If we manage to be able. Let's see, we we manage to to get the Lcf.

[Dirk Hufnagel] 11:50:04
Working, and we can run like, really, lo, super, large scale. Yeah, So we figured out, is this going to be okay?

[Dirk Hufnagel] 11:50:15
Because at that point it's going to be a much larger fraction of it.

[Dirk Hufnagel] 11:50:18
Might be a much larger fraction of overall resources Is this beyond pledge models still working, then, in this case.

[Paolo Calafiura (he)] 11:50:22
Hmm.

[Paolo Calafiura (he)] 11:50:29
I I I would say, we'll cross the bridge if we can to it.

[Paolo Calafiura (he)] 11:50:33
But you know, in India and the India it's I guess, What I'm saying is that I'm not sure it is a particularly important distinction right now.

[Paolo Calafiura (he)] 11:50:44
That resources pledges are not pledged, That's that's I guess what I'm trying to say.

[Paolo Calafiura (he)] 11:50:52
And yeah, you are right. If if we end up ending the 75% of our resources being the non pledge, then it's weird.

[Paolo Calafiura (he)] 11:51:00
Yes, Oh, we have in perfect

[Dirk Hufnagel] 11:51:01
I mean. And also Brian made a good argument on on Monday is at some point. So far we're looking at this from our viewpoint at some point it might become a problem for the agencies, because they want credit for it so that could become an issue

[Paolo Calafiura (he)] 11:51:21
Well, we do, at least, I'm sure you do the same, at least in Atlas we do.

[Paolo Calafiura (he)] 11:51:24
We do a knowledge with that content

[Dirk Hufnagel] 11:51:26
Yes, we reported, but but that, as far as Wcg.

[Dirk Hufnagel] 11:51:31
Is concerned they're not. I don't know. There's second class resources, so I don't know how much they've met us to the agency

[Paolo Calafiura (he)] 11:51:37
Well, I think it's far. It's part of one of these is concerned that Wcg.

[Paolo Calafiura (he)] 11:51:42
Does a monster

[Paolo Calafiura (he)] 11:51:47
yes.

[Dirk Hufnagel] 11:51:58
What about the the computing plans?

[Douglas Benjamin] 11:52:01
and can I ask them the question another way? In the next 5 years?

[Douglas Benjamin] 11:52:04
Will we need these to meet our pledge, Kevin.

[Douglas Benjamin] 11:52:10
Our current given sort of flat funding

[Douglas Benjamin] 11:52:17
Because you said a 5 year timeline

[Dirk Hufnagel] 11:52:20
Yeah, that's what I That's I had a similar thought, a question that basically because pledge pledge means something in terms of what you can plan.

[Dirk Hufnagel] 11:52:30
Right. I mean the you base your planning on on pledge and beyond pledge is something you could add extra.

[Enrico Fermi Institute] 11:52:30
Okay.

[Dirk Hufnagel] 11:52:37
So

[Dirk Hufnagel] 11:52:41
If that extra becomes required to to what you need to do as a baseline, doesn't, it need to be included in the pledge

[Enrico Fermi Institute] 11:53:02
okay.

[Dirk Hufnagel] 11:53:11
I guess no one has an answer for that.

[Enrico Fermi Institute] 11:53:19
Just need more time to think about

[Dirk Hufnagel] 11:53:21
Yeah, I mean, it's I don't think I mean, this is this is future.

[Dirk Hufnagel] 11:53:26
This is I think these are The questions should go into report.

[Dirk Hufnagel] 11:53:30
But it's it's and then it's it's anyway outside the scope of this workshop. And maybe even this report in general, these these kinda discussions on this topic.

[Douglas Benjamin] 11:53:42
But how much labor do we want, or for beyond pledge activity, how much labor is acceptable versus excessive

[Douglas Benjamin] 11:53:57
In other words, if if it takes 3 ftees to do 3% of the Monte Carlo Atlas needs us contribution, then you might consider that excessive

[Dirk Hufnagel] 11:54:10
Holy, you! You wanna weigh in on this

[Oliver Gutsche] 11:54:13
Well, I let let me try, so I I think the agencies are also trying to optimize their budget.

[Oliver Gutsche] 11:54:24
Right? So in the end that the agencies need to enable us to do our site.

[Oliver Gutsche] 11:54:30
So, if if the agencies have the possibility to say, okay, instead of getting you all the money to on all your sites, some of your processing will come from reliable allocations on Hpc: then the question is, do they fulfill the requirements to be

[Oliver Gutsche] 11:54:52
acknowledged as this official contribution to the experiment, and then it becomes a cost question right as as you asked, how many fts is reasonable.

[Oliver Gutsche] 11:55:05
It depends, then, how much money, how much funds you would actually save by this approach.

[Oliver Gutsche] 11:55:12
So So I think the question that we might have to answer is, How much does it cost us to pledge Hpc.

[Oliver Gutsche] 11:55:23
Resources, and that then goes into the calculation of us.

[Oliver Gutsche] 11:55:30
And the agencies of how use, and how how how efficient it is to actually pledge Hpc resources for our purposes.

[Enrico Fermi Institute] 11:55:32
Okay.

[Oliver Gutsche] 11:55:39
So that would be very interesting assessment for us.

[Enrico Fermi Institute] 11:55:45
Well, and and all of the

[Oliver Gutsche] 11:55:46
I don't know if I make Yeah, sure. Sorry. Go ahead.

[Enrico Fermi Institute] 11:55:50
No, I'm just gonna say that. But then the other half of that is, if if if and you're right, that we will have to answer the question at some point, how much does it cost to provide X resources from Hbc's but then we also need to be ready for the immediate other question which

[Enrico Fermi Institute] 11:56:03
is, going to be. How much does it cost for us to provide those same resources on premises?

[Enrico Fermi Institute] 11:56:09
I guess right.

[Oliver Gutsche] 11:56:11
yeah. But I mean for for the letter question, I mean, we we we have 15 years of experience to do that. Right?

[Oliver Gutsche] 11:56:21
So so so for me, for for for this this exercise.

[Oliver Gutsche] 11:56:26
It's really about. If we want to have Hpc resources, all commercial class resources replace pledges that we normally would provide through our sites.

[Oliver Gutsche] 11:56:37
What would it actually cost

[Paolo Calafiura (he)] 11:56:51
one thing, one thing which I guess I cannot do to this is that, and to clarify what I was saying before, is, that I think I think that what we need right now, and that's why that the new benchmarking is so important.

[Paolo Calafiura (he)] 11:57:06
What we need right now is A is a reliable way to do accounting.

[Paolo Calafiura (he)] 11:57:13
And so, for example, we can then answer the question. The tag was asking, Is it worth having?

[Paolo Calafiura (he)] 11:57:18
3 fps to get 3% of that of the all good resources I I don't want to use the word pledge, though nonplex with us is forget about that.

[Enrico Fermi Institute] 11:57:27
okay.

[Paolo Calafiura (he)] 11:57:28
If if I have, if a site is giving me 3% of the resources, how much money, how much effort, and therefore money do I need? That?

[Paolo Calafiura (he)] 11:57:38
Is it worth it? Put into it so that I think the the problem is that right now we do not.

[Paolo Calafiura (he)] 11:57:46
Well, we we set also the problem. That is for that we do not.

[Paolo Calafiura (he)] 11:57:52
We do not have basically any work. Folks running on on Lcs, because we don't have any accelerated workloads really in production.

[Paolo Calafiura (he)] 11:57:57
But the assuming we do add them, then we need that.

[Paolo Calafiura (he)] 11:58:01
Then we need the hey? Hey? Cool way to measure what is the contribution?

[Paolo Calafiura (he)] 11:58:08
Could be the number of events. But then, what? Of what? Oh, what kind, what kind of events is?

[Paolo Calafiura (he)] 11:58:15
That is it? The full simulation simulation, reconstruction.

[Paolo Calafiura (he)] 11:58:19
So it is for me that the critical question is the accounting is not the pledging.

[Paolo Calafiura (he)] 11:58:24
I'm not okay. Just to reformulate what I was saying before

[Enrico Fermi Institute] 11:58:40
okay.

[Dirk Hufnagel] 11:58:46
I couldn't do. We wanna move on. I don't think that's I mean we have many questions, but since it concerns the future as it's expected.

[Enrico Fermi Institute] 11:58:49
Yeah.

[Enrico Fermi Institute] 11:58:55
Need. Yeah, So the next slide was just placeholder for that.

[Dirk Hufnagel] 11:59:02
And then another question from the charge was, what new facility features the policies would help use.

[Dirk Hufnagel] 11:59:09
Atlas and Usc. Adopt Commercial Cloud and Hbc.

[Dirk Hufnagel] 11:59:13
Resources, one thing that we had in here was security Don't we invited some security folks, But I don't think anyone actually managed to connect.

[Dirk Hufnagel] 11:59:24
So that's a little bit of a pity. But one big problem.

[Dirk Hufnagel] 11:59:31
Why, apart from the elsef restriction with the no outbound Internet, from the worker notes also most of the Hpc these days have some saw of Alright Mfa login procedure, so you cannot really connect from the outside to to the to desk systems without going through some

[Enrico Fermi Institute] 11:59:37
Yeah.

[Dirk Hufnagel] 11:59:52
mfa process, and that usually would mean that we cannot really integrate things in an automated provisioning systems and things like that.

[Dirk Hufnagel] 12:00:01
But mfa some Hbc's a bit more flexible.

[Dirk Hufnagel] 12:00:06
What they allow Mfa to meet like at at Lcfs.

[Enrico Fermi Institute] 12:00:08
Yeah.

[Dirk Hufnagel] 12:00:12
It's basically strict hardware talking. So phone apps, so you can't do anything unless you and any kind of outside connection goes through that step.

[Dirk Hufnagel] 12:00:20
So cannot be automated. The Nsfunded Hbc.

[Dirk Hufnagel] 12:00:26
So far at least more forgiving. They'll they can say, Okay, M.

[Dirk Hufnagel] 12:00:30
If it can mean that that the system that locks in remotely is comes from a certain Ip or or or things like that or they allow Mfa to be bypassed at the moment, still in general as as a policy, question and then fernando you want to say something on

[Dirk Hufnagel] 12:00:47
the the cloud issues with the with the Cas

[Fernando Harald Barreiro Megino] 12:00:51
yeah. So they are. The problem. Is that so? The Oh, the group on Amazon.

[Fernando Harald Barreiro Megino] 12:01:01
They use the their own certificate authorities, and those are not trusted by Igtf.

[Fernando Harald Barreiro Megino] 12:01:11
And in particular, if you want to do a third party transfer, it means great spite.

[Fernando Harald Barreiro Megino] 12:01:15
The cloud, The cloud ca is not trusted, and then the transfer phase, and you need to do some.

[Fernando Harald Barreiro Megino] 12:01:23
I put something in front of the this storage with another certificate, and this is it it?

[Fernando Harald Barreiro Megino] 12:01:34
It it can become a bottleneck. It could be preferable if if the third party transfers would work, This is being discussed already in the Wcg.

[Fernando Harald Barreiro Megino] 12:01:47
But I heard that they that probably there will not be a solution in the next.

[Fernando Harald Barreiro Megino] 12:01:55
Yes, but this is some to long term problem

[Enrico Fermi Institute] 12:02:02
So? Is the is the issue that that the the Wcg.

[Enrico Fermi Institute] 12:02:05
Would need to accept the the certificate authorities of the of the Commercial Cloud Providers

[Fernando Harald Barreiro Megino] 12:02:12
Yes, they would have to become part of the ittf, and I don't know exactly what are the the policies to get into the igts, and I understand it also require some effort from the from the Cloud Button for them.

[Fernando Harald Barreiro Megino] 12:02:31
Maybe it's not. It's not worth it. So that's why there is.

[Fernando Harald Barreiro Megino] 12:02:37
There is not really a solution. This will tell me, as far as I understand.

[Enrico Fermi Institute] 12:02:43
I'll just point out that it's but the ittf is is bigger than just the Wcg.

[Enrico Fermi Institute] 12:02:47
So it's not necessarily Wcg. Has to. It's a It's a step above them that we would have to convincing to do it

[Dirk Hufnagel] 12:03:02
Oh! And that I just wanted to. I skipped the Federated Identity point because I think, without a security person from Lap here, there's no point discussing that I I just to to mention what it is before I to Eric that basically, the labs have been working on, kind of

[Dirk Hufnagel] 12:03:24
federating the systems in terms of logins, and and so on.

[Dirk Hufnagel] 12:03:28
but I'm not sure if this will help, because I, as far as I know, Mfa, is still required on top of that, so I might be able to log into an to argon with my firmelap id but I still would have to go for the mfa step as far as I know I I just don't

[Dirk Hufnagel] 12:03:44
know if that's something that eventually could dropped with some, maybe private networks between the the the the national apps.

[Dirk Hufnagel] 12:03:51
But that's I wanted to get seek back from it.

[Dirk Hufnagel] 12:03:54
Security. I guess we have to offline Eric

[Eric Lancon] 12:03:58
yes, I wanted to comment. I have 2 points. So when you come to me fair identification and cloud cas, that's a price to pay when you don't own the resources.

[Eric Lancon] 12:04:16
So instead of complaining, we should find innovative solution whenever it's possible to use better those solutions on the Federated it.

[Eric Lancon] 12:04:30
I I invited the people online for this meeting.

[Eric Lancon] 12:04:35
That's so. If there are no the feminine people, or we can have a the Dnn perspective Your home is on the line wanted to to remind that me Fair will certainly become a Stand up everywhere.

[Eric Lancon] 12:04:56
Well, it's a 5 years time scale, so we should adapt to and foresee that we'd have to work with it.

[Eric Lancon] 12:05:06
So on, federated id If you want to say a few words, but

[Jerome Lauret (he/him)] 12:05:10
cool, sure. Let Let Let me write away. Jump in the nfa topic and close.

[Jerome Lauret (he/him)] 12:05:19
This in general, any services that we have put in place, that are cloud inspired, or port provide access for a wide variety of people from multiple different organization.

[Jerome Lauret (he/him)] 12:05:35
Required. Mfa. There is. There's been no no escape to it So far we have been able to set up.

[Jerome Lauret (he/him)] 12:05:43
Of course you know a few services, and Gp. To Herb, and things like that.

[Jerome Lauret (he/him)] 12:05:47
But has been essentially the prerequisite.

[Jerome Lauret (he/him)] 12:05:53
do you the other comment on federated Id is that, of course it all depends on the Federation, and we have been basically somewhat allowed to proceed with many of the Trusted Federation.

[Jerome Lauret (he/him)] 12:06:10
Of what Dewey sees it as trusted. So that is, for example, people coming from all the national labs.

[Jerome Lauret (he/him)] 12:06:18
we have an exemption for certain federation as well.

[Jerome Lauret (he/him)] 12:06:23
but in general, for example, we have had a consistent message that Google, for example, is not a trusted and acceptable federation and the reason for that is just that anybody anytime can impersonate anyone And and a live here on our side to we saw something kind of funny funny identity so

[Jerome Lauret (he/him)] 12:06:47
that's all I wanted to say. But of course there's a lot of Paul guys already seeing the the fact that we are being told.

[Jerome Lauret (he/him)] 12:06:55
Okay, please. Proceed with directed, Id is very encouraging. But this is a long road, and I think that early on someone mentioned also that you know the and Mfa.

[Jerome Lauret (he/him)] 12:07:09
Or Sfa could be bypassed in some ways.

[Jerome Lauret (he/him)] 12:07:10
That's yes, that's true. And there is a lot of work here.

[Enrico Fermi Institute] 12:07:12
Okay.

[Jerome Lauret (he/him)] 12:07:13
to also add trusted metadata. You know, as part of the certificate; but this is indeed not an immediate E immediate, development.

[Jerome Lauret (he/him)] 12:07:24
So that's why, perhaps right now do we prefer to have only trusted Federation, just to be sure that you know everybody is essentially the same kind of whole of engagement

[Enrico Fermi Institute] 12:07:40
So? There. 1 One question I had is, you know, security perspective.

[Enrico Fermi Institute] 12:07:45
Is is it? Is it the case that Mfa is fundamentally analogous to the that you want to have some human interaction to authenticity with a resource, or you know does the mfa.

[Enrico Fermi Institute] 12:08:04
Just mean that you you know, you really just need these multiple facts, right?

[Enrico Fermi Institute] 12:08:09
It's it's not sufficient just to have a key or password, or whatever you need to have some additional factor

[Jerome Lauret (he/him)] 12:08:15
Cool. Cool. Go back. It's a second. It's a simple, and in fact, you know in some cases well, even required to have a secret handshake.

[Jerome Lauret (he/him)] 12:08:27
I mean, okay, essentially, know the point of contact from within the experiment that you are in order to have an account, and being approved because this is the level of confidence depends, of course, on the service that you access just yeah, just to be sure that you you you see the difference.

[Enrico Fermi Institute] 12:08:30
And

[Enrico Fermi Institute] 12:08:35
Sure.

[Enrico Fermi Institute] 12:08:41
Sure.

[Jerome Lauret (he/him)] 12:08:46
Is that if you access matter most, for example and book, even your federated Id is enough.

[Jerome Lauret (he/him)] 12:08:53
If you are issue accounts something that essentially allows you to make modification to the content.

[Jerome Lauret (he/him)] 12:09:02
Then me phase require you, and if you do access computing resource right now, things where you can eventually launch a large number of jobs that so the knee fumble Kevin, that's in the up here as being some kind of illegal activity then not only it's an but in order to have your account

[Jerome Lauret (he/him)] 12:09:20
approved. You need some extra steps and verification that we came up with a procedure that was actually acceptable by outside the team.

[Jerome Lauret (he/him)] 12:09:29
but you know, so there's some kind of like jail of acceptance

[Enrico Fermi Institute] 12:09:35
So so I guess then, is it fair to say that Mfa.

[Enrico Fermi Institute] 12:09:40
Does not. Fundamentally, you know, schoolude automation

[Jerome Lauret (he/him)] 12:09:47
I would say

[Enrico Fermi Institute] 12:09:47
Renton what you know that you know, For example, Tak has done where they where they consider, a you know, a trusted machine to be a factor on top of you know the the key that you provide

[Jerome Lauret (he/him)] 12:10:01
Right. So actually, actually, this is a excellent question, because, of course, for example, jobs, submission from a trusted host.

[Jerome Lauret (he/him)] 12:10:11
Right, for example, in Osg Land has been, of course, accepted right so.

[Enrico Fermi Institute] 12:10:16
Okay.

[Jerome Lauret (he/him)] 12:10:17
And that indeed, is somewhat what you are hinting, as you know, that host is trusted, and you know to access that host to submit.

[Jerome Lauret (he/him)] 12:10:30
Then you require, Then you have, additional you know, authentication.

[Jerome Lauret (he/him)] 12:10:35
You understand what I'm saying right? I mean you you Log, for example, to that host using your local credential.

[Jerome Lauret (he/him)] 12:10:40
Then you eventually issue a token, or whatever which is yet a second factor of then use to meet your job, and that that has been accepted for quite a while.

[Jerome Lauret (he/him)] 12:10:51
So you you You are right, that there may be some leeway there, some some home in in that sense.

[Jerome Lauret (he/him)] 12:10:58
Okay.

[Enrico Fermi Institute] 12:11:00
I think

[Enrico Fermi Institute] 12:11:09
Other other comments or questions about mfa or Cloud. Ca's httf sort of thing

[Dale Carder] 12:11:13
I know a nurse. There. There's a process to get long term keys instead of like the default.

[Dale Carder] 12:11:18
24 h key for Ssh. Proxy

[Dirk Hufnagel] 12:11:21
Yeah, it's it's up to a month, I think, to support

[Dale Carder] 12:11:24
Yeah.

[Dirk Hufnagel] 12:11:27
But you do need to get that key. You need to go through an mfa process, and then you're okay.

[Enrico Fermi Institute] 12:11:34
Sure.

[Dirk Hufnagel] 12:11:34
For 30 days. They They basically that's that's the compromise between.

[Dirk Hufnagel] 12:11:37
We don't want to allow automation when no one authenticates for a couple of years, and then an age will.

[Dirk Hufnagel] 12:11:42
If the that key gets compromised, then basically everyone can use it forever to you. At least you don't have to do this meeting, at least operational feasible.

[Dirk Hufnagel] 12:11:53
To go to to use, to use the system, even with the Mfa. Rules.

[Dirk Hufnagel] 12:11:57
Some place.

[Dirk Hufnagel] 12:12:01
And I mean globus online, is the same. That's what we're doing with the the rush. You online.

[Dirk Hufnagel] 12:12:06
Integration for the transfers, someone actually has to log in manually to the portal and renewed a key once a week so that the transfers can keep keep going But it's it's I mean once a week, once a month that's all that just means you roll it into cost it's

[Enrico Fermi Institute] 12:12:15
Right.

[Dirk Hufnagel] 12:12:23
it's it's it's a bump on the cost for the operations for the long term.

[Dirk Hufnagel] 12:12:26
Information

[Jerome Lauret (he/him)] 12:12:29
usually when you have those long term credentials, so you also have to demonstrate that you have a way to revoke it first.

[Dirk Hufnagel] 12:12:41
I don't know how to handle it. You would probably have to go through nurse, because I don't think the you can revoke it yourself.

[Jerome Lauret (he/him)] 12:12:48
yeah, exactly. So this may be a concern in a long term.

[Jerome Lauret (he/him)] 12:12:52
Yes, I'm just saying, in terms of visibility, people may not know the detail, but usually that's one of the things that would come when a long credentials appeal

[Robert Hancock] 12:13:05
yeah, and in our plan, with the vote, right? So the long-standing credentials would state the Volt.

[Robert Hancock] 12:13:11
Server, so we could just delete them from there, and then they wouldn't be able to pull any more short term credentials, you know.

[Robert Hancock] 12:13:15
Short term tokens, we access token

[Enrico Fermi Institute] 12:13:27
Okay.

[Dirk Hufnagel] 12:13:31
if you don't have any more comments on security topics, it could me move on to the allocations and that's more like acquiring the resources.

[Enrico Fermi Institute] 12:13:42
Okay.

[Dirk Hufnagel] 12:13:45
So an Hbc: you do it through. It's currently yearly.

[Dirk Hufnagel] 12:13:48
Our locations. So we are. It was mentioned already in the Hbc. Focus area discussions.

[Dirk Hufnagel] 12:13:55
This: if you had multi year locations, that would you reduce that, hey?

[Enrico Fermi Institute] 12:13:55
Okay.

[Dirk Hufnagel] 12:14:00
Reduce the effort to acquire the Hbc resources because you wouldn't have to constantly rejustified every year and B.

[Dirk Hufnagel] 12:14:09
It would also open up possibilities to include sizable Hbc.

[Dirk Hufnagel] 12:14:15
allocations in the in the planning process which you can do right now, because at the moment you write the proposal, you get the decision, and then, usually on the order of few months, later 1, 1, 2, few months later, you get that you actually have the resources, and you don't actually get the decision, until a few

[Dirk Hufnagel] 12:14:37
months before which is too late to actually include it in a name of the long term, in in in the long term, planning planning process for research use in the experiments, and the that's that's from that side independently of of any kind of pledging problems we have that's that's

[Dirk Hufnagel] 12:14:55
a That's a problem in in and being able to pledge.

[Dirk Hufnagel] 12:14:58
I mean, if you don't know that we have the resources we cannot pledge it, even if there would be procedures in place to be able to do so technically, and then what was mentioned also many times before is that a large stored allocations with connectivity to the white area network would

[Dirk Hufnagel] 12:15:14
allow, would simplify Hbc operations, which was basically would make some sinks possible.

[Dirk Hufnagel] 12:15:23
that might not be possible now, and it definitely would reduce the cost

[Enrico Fermi Institute] 12:15:29
Yeah, okay.

[Dirk Hufnagel] 12:15:30
And on the cloud side of Anando. You want to say something on the subscription model

[Fernando Harald Barreiro Megino] 12:15:37
well, I mean I'm not sure exactly how thing works.

[Fernando Harald Barreiro Megino] 12:15:44
At the end. The cloud, that's as long as they they get the check on the on the subscription is renewed, and also the there are come on like there is a common understanding out what is going to be the cost of the subscription.

[Fernando Harald Barreiro Megino] 12:16:02
It's okay. But then I don't know in in in Atlas how how the budgeting works will prepare those a yearly parts it for that

[Dirk Hufnagel] 12:16:18
But I'm really curious about what is what will happen after the it what is it?

[Dirk Hufnagel] 12:16:24
Fifth, 15 months.

[Fernando Harald Barreiro Megino] 12:16:25
Yes, it's around October 2023.

[Fernando Harald Barreiro Megino] 12:16:29
That's fine.

[Dirk Hufnagel] 12:16:30
We'll we'll see what I mean. I really would like to see what happens.

[Dirk Hufnagel] 12:16:32
Sand. If they just renew it at the same, or if they actually drilling into the into the billing data that they collect and and do some, I mean, it depends, I guess, on the billing data.

[Enrico Fermi Institute] 12:16:40
Okay.

[Dirk Hufnagel] 12:16:46
But but still I I'm curious.

[Dirk Hufnagel] 12:16:52
And then some specific topics, Here, on the Hpc.

[Dirk Hufnagel] 12:16:56
Side we we mentioned that the facilitating Cbm Fs access.

[Dirk Hufnagel] 12:17:01
I think this is small as a self problem, because see what as as Brian said, if excess is, is considered kind of stable, and the solution to provide.

[Enrico Fermi Institute] 12:17:03
Okay.

[Dirk Hufnagel] 12:17:12
Cdfs these days. So in that that basically immediately gets you to the second problem, you need to have the ability to or either have a some squid infrastructure in place, or the ability for to launch our own because that supports Cdm Fs exec And then there's frontier on top of it but

[Dirk Hufnagel] 12:17:33
the first of all access at facilities access to software. Oh, is it as a comment

[John Steven De Stefano Jr] 12:17:42
Yeah, I was just wondering in general, on the Hpc side when it comes to Cbn Fs access.

[John Steven De Stefano Jr] 12:17:48
What the the mean issue is with the native client. Is it just connectivity that's restricted on the

[Dirk Hufnagel] 12:17:54
It's usually that they don't want to install custom software for just one customer

[John Steven De Stefano Jr] 12:17:59
But they'll, and they'll use Cbm physics sick

[Dirk Hufnagel] 12:18:02
Well, that runs completely in user space. The latest versions It's it's becoming increasing to I mean, I've reached.

[John Steven De Stefano Jr] 12:18:05
True

[Dirk Hufnagel] 12:18:10
We've started using it like a year or 2 ago, like 2 years ago, and it's becoming increasingly easier to use it because the newer machines run new operating systems with newer kernel features.

[Enrico Fermi Institute] 12:18:16
Yeah.

[Dirk Hufnagel] 12:18:21
And it's basically with this at this level, you can run it completely in user space.

[Dirk Hufnagel] 12:18:27
You don't you? Really? The system dependencies are so small these days. If the kernel is new enough that that it kind of it just works

[Enrico Fermi Institute] 12:18:36
In in the past at least, you know, justified or not.

[Enrico Fermi Institute] 12:18:39
There there was certainly some paranoia I I have seen about, you know, running fuse file systems on a compute note.

[Enrico Fermi Institute] 12:18:48
some some sites were worried about that. Like doesn't have any

[Dirk Hufnagel] 12:18:53
Yeah, yeah, some sites, yeah, some some. Hpc: size, you log into a batch node, and like fuse amount, is not available.

[Dirk Hufnagel] 12:19:00
But that's not a If the kernel is new enough.

[Dirk Hufnagel] 12:19:02
See me, Xxx doesn't need fuse amount binary to do a fuse mount.

[Dirk Hufnagel] 12:19:07
You can do it directly through Yeah.

[John Steven De Stefano Jr] 12:19:11
Sure, and I understand the concern about views being another layer on top of, and already simplified or complex system.

[John Steven De Stefano Jr] 12:19:18
But I think the native client has proven fairly stable lately, so I understand the concerns.

[Enrico Fermi Institute] 12:19:24
So it's just it's just convincing the sites.

[Enrico Fermi Institute] 12:19:26
That's the case. Okay.

[John Steven De Stefano Jr] 12:19:27
Thanks.

[John Steven De Stefano Jr] 12:19:27
Yeah.

[Dirk Hufnagel] 12:19:33
And then another area. The Hbc. If you have it, makes so much simpler, if you, if they would provide right, rushio compatible storage we are we're currently working with with nurse on that Lcf: I don't think I mean there's there's no efforts

[Dirk Hufnagel] 12:19:51
there, and I I'm not sure if it will ever happen.

[Dirk Hufnagel] 12:19:54
But at least they support globus online. So we do have a grocery club gloves online integration.

[Dirk Hufnagel] 12:20:01
So it's it's doable

[Douglas Benjamin] 12:20:05
So then we just have to make sure that we call out that there's a hop that's required

[Dirk Hufnagel] 12:20:12
Yeah, I mean, we. We tried. I don't know if you ever tried.

[Dirk Hufnagel] 12:20:15
The multi hop. We We tried it through, nurse, and it just worked

[Dirk Hufnagel] 12:20:21
So we had nurse. Currently We have nurse currently integrated via the still existing good Ftp.

[Dirk Hufnagel] 12:20:25
Integration which will eventually go away. But it's still there, for now and then both nurse and hey, I'll see a data where integrated into be a global. Online. And then basically it's it's you can configure.

[Dirk Hufnagel] 12:20:41
Configure the roo system. So that when you, when you put in a rule, we create some data, theta it automatically. First, that's a good ftp trap transfer to nurse and then immediately global online transfer from nurse to theta

[Douglas Benjamin] 12:21:02
but will that work? When nurse goes to the next generation, or globus

[Dirk Hufnagel] 12:21:07
No, but that's that's what the work with nurse, where the work with nurse on the d interface is important, because that eventually hopefully will replace the good fft integration which which is deprecated since many years and will go away

[Douglas Benjamin] 12:21:28
So Sam, us is planning to keep nurse in the Hpc data flow path for essentially Nsf.

[Douglas Benjamin] 12:21:39
And other doe. Hpcs

[Douglas Benjamin] 12:21:44
Versus putting Fermi lab in the path, so that Fermi lab becomes the connector

[Dirk Hufnagel] 12:21:45
Yes, I'm

[Dirk Hufnagel] 12:21:53
where we exactly put the multi hop is still to be decided, and ask, is is an obvious candidate.

[Dirk Hufnagel] 12:21:58
But we also, I think we have tattoos with global online licenses.

[Dirk Hufnagel] 12:22:02
So that would be some that would be an alternative option that we have

[Douglas Benjamin] 12:22:11
because I have less uses, you know, as the Hop

[Dirk Hufnagel] 12:22:14
Yeah.

[Dirk Hufnagel] 12:22:23
I mean the at the moment, at the level of transfers, we need to do to the to the Lcf.

[Enrico Fermi Institute] 12:22:29
See.

[Dirk Hufnagel] 12:22:30
It's it's not that important where the hop location is.

[Dirk Hufnagel] 12:22:33
If we scale up Lcf Usage and we really looking at the future.

[Dirk Hufnagel] 12:22:38
That's like heavy on like data reconstruction, or so then it becomes, and more important question, because that's potentially a lot of traffic you have to multi-hop

[Enrico Fermi Institute] 12:22:54
Okay.

[Dirk Hufnagel] 12:23:08
and then we had a point here that we on on network traffic and pairings to improve connectivity and reduce the limited e rate cost.

[Dirk Hufnagel] 12:23:18
I think we had an interesting presentation from me, as not yesterday, about the connectivity side of things.

[Dirk Hufnagel] 12:23:22
I don't think we got anywhere with the reduce unlimited egret cost.

[Dirk Hufnagel] 12:23:26
That's kinda that's more. That's not so much a question of the how this, how the the networks are connected, and what peering how the peering is set up but more question of what type of cost model you have you have a subscription or you use a cloud that doesn't

[Dirk Hufnagel] 12:23:42
have egret. Yeah, it seems that's the outcome I get out of this Berkshire

[Dirk Hufnagel] 12:23:50
And then there's an open anded question, What else is there anything?

[Dirk Hufnagel] 12:23:54
We forgot to cover here that that could help with the without our Hbc.

[Enrico Fermi Institute] 12:24:12
and I think that's that's what we have for this session.

[Enrico Fermi Institute] 12:24:18
Call current standards.

[Enrico Fermi Institute] 12:24:23
Yeah, So you know, if there are other things that we that we should talk about, for you know, facility, features, and policies, or you know any of the the topics that we covered in previous days I think it, would be a good time, to bring them up now

[Enrico Fermi Institute] 12:24:44
otherwise we can. We can go about the session a little early

[Paolo Calafiura (he)] 12:24:54
when do we reconnect? If we

[Enrico Fermi Institute] 12:24:57
the next session will be at at one o'clock central time.

[Paolo Calafiura (he)] 12:25:03
Okay.

[Enrico Fermi Institute] 12:25:17
seeing people disconnect. So maybe we'll just go ahead and and and close out, and then resume in an hour and a half

[Dirk Hufnagel] 12:25:23
sounds good

[Enrico Fermi Institute] 12:25:24
Okay, So it's guessing

[Paolo Calafiura (he)] 12:25:24
hi folks.

[Fernando Harald Barreiro Megino] 12:25:25
okay.