Pantheon and Puffer

[Halimat Adetutu]

Hi, 

I'm a masters research student and I came across pantheon and puffer. Out of curiosity, is there any relationship between pantheon and puffer or they two standalone entities.

regards,

[Keith Winstein]

Hello Halimat -- I'd say Pantheon probably inspired us to do Puffer.

Pantheon is (1) a collection of congestion-control schemes that all expose the same interface, (2) a [now dormant] collection of roughly 15 nodes around the world, many with connections to local cellular networks, and (3) a data archive of head-to-head round-robin trials of the various congestion-control schemes running in various scenarios (single flow, multiple flows) over (a) all-wired and partly-wireless paths between the nodes around the world, (b) emulation scenarios calibrated to match part (a), and (c) "torture test" emulation scenarios, as well as (4) this set of emulator configurations [(b) and (c)] that anybody can run their congestion-control schemes over.

We learned a lot from doing Pantheon (some of which was about the dangers of putting too much trust in results from network emulation). Puffer is a much bigger but less-controlled project that runs head-to-head trials of ABR, congestion-control, and network-prediction schemes over hundreds of thousands of paths (between our one node, and all the Puffer viewers) and reports the results every day. Puffer probably collects more real-world data in a day than Pantheon ever did, and the real-world data is a lot more representative of real Internet paths.

Hope this is helpful, and best regards,

Keith

--
You received this message because you are subscribed to the Google Groups "Pantheon" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pantheon-stanf...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pantheon-stanford/e26ad767-740f-4531-a6b7-444c2a25b90dn%40googlegroups.com.

[Halimat Adetutu]

Yes. Your response was helpful!

thank you,