Thanks
", "time": "2023-03-29T06:50:16Z"}, {"author": "Adrian Farrel", "text": "To Dean's point: it is really important that you all contribute to the on-list discussion saying which use cases you think are particularly \"realistic\" and need to find their way into the use case draft. You may be asked to contribute text!
", "time": "2023-03-29T07:07:45Z"}, {"author": "Luis Contreras", "text": "Regarding metrics, it would be good to have coordination between CATS and ALTO, since there is some on-going propositions in ALTO to consider compute- and service-function related metrics, as well
", "time": "2023-03-29T07:19:00Z"}, {"author": "Adrian Farrel", "text": "Yes, Luis, we've come to that conclusion, too. Hang on for presentation 6a.
", "time": "2023-03-29T07:21:07Z"}, {"author": "Adrian Farrel", "text": "Although ALTO and CATS solve a different architecture, they very probably operate on the same/similar set of metrics.
", "time": "2023-03-29T07:22:46Z"}, {"author": "Adrian Farrel", "text": "Further, the component that captures the metrics could be the same. Unclear whether the metric distribution mechanism is the same (to CATS router or to ALTO server)
", "time": "2023-03-29T07:23:34Z"}, {"author": "Mohamed Boucadair", "text": "One comment about slides 3/4, please note that the title indicates (current) to express that the arch is not forzen and that this is right a starting point.
", "time": "2023-03-29T07:24:18Z"}, {"author": "Jan Lindblad", "text": "Speaking of metrics, a class of metrics I find missing in all these presentations (and in ALTO) are metrics for cost of energy consumption, and CO2 intensity. In my world, that is a metric that service providers think is important today, and increasingly important for the future. This class of metrics should fit in well in the existing architecture, I believe.
", "time": "2023-03-29T07:36:30Z"}, {"author": "Hang Shi", "text": "Do you think cost of energy will vary between sites?
", "time": "2023-03-29T07:37:15Z"}, {"author": "Adrian Farrel", "text": "Yes, Jan. But do you think this should be folded in here, or more generally into routing work?
", "time": "2023-03-29T07:37:45Z"}, {"author": "Jan Lindblad", "text": "Definitely. Cost of energy (and CO2 footprint) varies greatly between energy providers.
", "time": "2023-03-29T07:37:50Z"}, {"author": "Adrian Farrel", "text": "Hang, I believe it varies very significantly. But also, different servers may have different energy consumption
", "time": "2023-03-29T07:38:26Z"}, {"author": "Jan Lindblad", "text": "In ALTO there is already a cost model in YANG. There are identities for many types of cost (latency, bw, .....) but not for cost of energy or CO2.
", "time": "2023-03-29T07:38:30Z"}, {"author": "Jan Lindblad", "text": "Probably makes sense to work with ALTO on those models. IMO we should then also add energy, CO2 as possible basis for selecting compute and network resources.
", "time": "2023-03-29T07:41:20Z"}, {"author": "Adrian Farrel", "text": "Alright. That was a commitment to that the CATS chairs will talk with the ALTO chairs about how to arrive (if appropriate) at a common set of computing metrics for use in both cases
", "time": "2023-03-29T07:45:20Z"}, {"author": "Mohamed Boucadair", "text": "An immediate action I see on the ALTO side is to make sure that the current cost groupings in the ALTO models are not too specific to ALTO and see how/whether to readjust so that this can be easily reused by CATS, and others.
", "time": "2023-03-29T07:48:46Z"}, {"author": "Jordi Ros-Giralt", "text": "Jan, that's correct, in some ALTO meetings the members have discussed the possibility of incorporating energy metrics, as that's increasingly important. We are currently going through these discussions and we will be having a few interim meetings in the next weeks (https://datatracker.ietf.org/wg/alto/meetings/), I would encourage your participation (drop a note to the ALTO mailing list on this topic)
", "time": "2023-03-29T07:51:16Z"}, {"author": "Boris Khasanov", "text": "Is BGP-LS a right approach for metric distribution here? Looks like some kind of subscription model fits better.
", "time": "2023-03-29T07:52:01Z"}, {"author": "Joel Halpern", "text": "I would prefer to have the requirements before picking protocols.
", "time": "2023-03-29T07:54:54Z"}, {"author": "Jan Lindblad", "text": "Thanks @Jordi, good to hear. Thank for the link and tip.
", "time": "2023-03-29T07:54:59Z"}, {"author": "Jordi Ros-Giralt", "text": "Med, let's address the alignment of ALTO metrics with CATs in the next interim meetings. I will put it in the agenda.
", "time": "2023-03-29T07:55:00Z"}, {"author": "Mohamed Boucadair", "text": "Looks like a good plan. You may refer to https://github.com/ietf-wg-alto/draft-ietf-alto-oam-yang/issues/34
", "time": "2023-03-29T07:55:51Z"}]