+1 on thanks to david
", "time": "2023-03-30T00:52:35Z"}, {"author": "J\u00e1nos Farkas", "text": "Thank you very much David!
", "time": "2023-03-30T00:52:40Z"}, {"author": "Greg Mirsky", "text": "As the saying goes \"You cannot control, what you cannot measure\" :slightly_smiling_face:
", "time": "2023-03-30T01:00:43Z"}, {"author": "David Black", "text": "There's at least one proposed solution that does that piecewise rather than end-to-end.
", "time": "2023-03-30T01:01:37Z"}, {"author": "Greg Mirsky", "text": "Sure, let's discuss it. The problem is that not all metrics are composable, i.e., the e2e metric can be composed of spatial metrics.
", "time": "2023-03-30T01:14:21Z"}, {"author": "David Black", "text": "This (ADN) is a proposed solution that does it piecewise (notice the large sigma in the center of the slide).
", "time": "2023-03-30T01:18:25Z"}, {"author": "Greg Mirsky", "text": "If anything allows for a delay variation, simple summarization of latencies may lead to incorrect conclusions
", "time": "2023-03-30T01:23:18Z"}, {"author": "Greg Mirsky", "text": "What is BW utilization of the proposed method?
", "time": "2023-03-30T01:25:05Z"}, {"author": "David Black", "text": "Sum is of upper bounds.
", "time": "2023-03-30T01:25:12Z"}, {"author": "David Black", "text": "On BW utilization, suggest asking Jinoo directly.
", "time": "2023-03-30T01:26:20Z"}, {"author": "Greg Mirsky", "text": "That might be acceptable estimate if delay variation is significantly smaller compared to the delay. Otherwise, we may end with overestimating the delay and, as a result, using too much of the resources.
", "time": "2023-03-30T01:27:41Z"}]