Huawei Cloud reveals the dynamic traffic allocation system it uses to cut bandwidth bills

Huawei has released details of how it manages its own cloud with a dynamic traffic allocation system optimized by machine learning and developed in response to surging demand for its services during the COVID-19 pandemic. The system was engineered using an array of operations research methodologies – such as continuous …

[Pascal Monett]

10 times the peak bandwidth ?

I think that's going to open some eyes at Redmond (and elsewhere).

Who knows ? Maybe Huawei will accept meeting with Western engineers and sharing the technique ?

[Roland6]

Re: 10 times the peak bandwidth ?

> Maybe Huawei will accept meeting with Western engineers and sharing the technique ?

That would mean admitting Huawei is ahead of the US; can’t see that sitting comfortably with the MAGA crowd and their rose tinted views on American tech.

As for sharing the technique, the simplest will be to purchase some Huawei networking equipment; which the MAGA crowd have been doing their best to prevent people from doing…

[Anonymous Coward]

Have to see more details to be sure, but credit where it's due

One of the big problems in that space is efficient use of resources, and maximizing parallel utilization. So an edge optimizer speculatively puts needed resources in play to ensure the massive capacity of edge links isn't idling due to core congestion.

Cloudflare has been doing much the same, but from the broad description the methods and results seem plausible. The Huawei solution leans into the math a bit, but while the specifics differ, the need and results are the same. Building a proper scheduler is tricky work, but pays out in a big way at scale. I'm not losing sleep over the prospects of this thawing US/China relations or the ongoing economic warfare, especially in the tech sector.