Hi network-health@/tor-relays@ mailing lists,

I noticed one thing: Tor relays on the West Coast US (and Canada) are generally slower than those on say the East Coast and in Europe.

I moved to the West Coast this January, but this was not an issue in the past when looking at dedicated servers I had from the West Coast prior to this year.

These are two middle relay instances on an Gigabit FTTH connection (Wave G in Redmond, WA):

https://metrics.torproject.org/rs.html#details/B0F9BA27944FA59E3B1A182208FF7... https://metrics.torproject.org/rs.html#details/DB710B14D7329B7289CFCC547F48E...

The Consensus Weight often is slower than the Advertised Bandwidth, and isn't ramping up despite having lots of bandwidth. I set RelayBandwidthRate to about 500 Mbps on each instance.

This isn't just a Wave G problem, this affects almost every relay in the West Coast including other Gigabit ISPs such as AT&T (AS7018) and Sonic (AS46375), as well as hosting companies and colocation facilities.

For instance, my Los Angeles-based Exit relays at Psychz Networks (AS40676) show this issue (however they are new):

https://metrics.torproject.org/rs.html#details/156AAC3FAD1ACC8906316519DCB44... https://metrics.torproject.org/rs.html#details/A69CEB30328B1E85C6B167FECAF2F...

Even Emerald Onion has this issue:

https://metrics.torproject.org/rs.html#details/09DCA3360179C6C8A5A20DDDE1C54...

What is causing this issue and is there a solution? Is it backbone congestion due to COVID-19? The high load on dirauths? sbws regressions?

Can I help fix this issue? I am a Core Tor contributor and am open to also working on sbws.

Is there a way to optimize my relays (they run FreeBSD).

-Neel

===

https://www.neelc.org/