Good evening fellow relay operators,
I haven't ever taken the time to configure bandwidth limits in torrc, always preferring to manage it at the firewall as we have other bandwidth limits set there as well. However, I'm curious - what do other relay operators prefer?
Make your day great, Isaac Grover, Senior I.T. Consultant Aileron I.T. - " Because #ProactiveIsBetter "
O: 715-377-0440tel:715-425-0440, F:715-690-1029tel:715-690-1029, W: www.aileronit.comhttp://www.aileronit.com/
LinkedIn: https://www.linkedin.com/in/IsaacGrover/ YouTube: https://www.youtube.com/channel/UCqrwZNFKdR-guKtuQzFPObQ
I usually do it in torrc, but am experimenting with letting my ISP/host (Verizon FiOS and GTHost.com) traffic shaping do this on unmetered connections.
The reason for this is because I learned about bandwidth accounting as mentioned on an earlier post here:
https://lists.torproject.org/pipermail/tor-relays/2018-December/016780.html (https://lists.torproject.org/pipermail/tor-relays/2018-December/016780.html)
Tor spends resources doing bandwidth accounting if you set RelayBandwidthRate and RelayBandwidthBurst.
So the runoff is: torrc is convenient, but doing it externally can offload the shaping and free the Tor process to do other things (after all, Tor is still singlethreaded) but adds complexity to your OS/network setup.
-Neel
===
January 14, 2019 10:09 PM, "Isaac Grover, Aileron I.T." <igrover@aileronit.com (mailto:igrover@aileronit.com?to=%22Isaac%20Grover,%20Aileron%20I.T.%22%20igrover@aileronit.com)> wrote: Good evening fellow relay operators,
I haven’t ever taken the time to configure bandwidth limits in torrc, always preferring to manage it at the firewall as we have other bandwidth limits set there as well. However, I’m curious - what do other relay operators prefer?
Make your day great, Isaac Grover, Senior I.T. Consultant Aileron I.T. – “ Because #ProactiveIsBetter ”
O: 715-377-0440 (javascript:false), F:715-690-1029 (javascript:false), W: www.aileronit.com (http://www.aileronit.com/)
LinkedIn: https://www.linkedin.com/in/IsaacGrover/ (https://www.linkedin.com/in/IsaacGrover/)
YouTube: https://www.youtube.com/channel/UCqrwZNFKdR-guKtuQzFPObQ (https://www.youtube.com/channel/UCqrwZNFKdR-guKtuQzFPObQ)
On 15 Jan 2019, at 13:09, Isaac Grover, Aileron I.T. igrover@aileronit.com wrote:
I haven’t ever taken the time to configure bandwidth limits in torrc, always preferring to manage it at the firewall as we have other bandwidth limits set there as well. However, I’m curious - what do other relay operators prefer?
How does your network handle packets that are over the limit?
Tor delays sending cells that would exceed its bandwidth limits. This increases in-process latency, but does not require network retransmits. But it also does not change the relevant TCP bandwidth delay product.
Most networks drop packets that are over their bandwidth limits. This increases network retransmits, and triggers a bunch of TCP algorithms that slow down the connection.
T
tor-relays@lists.torproject.org