On 2013-09-26 15:19, Moritz Bartl wrote:
On 09/26/2013 03:11 PM, phoneyball@vfemail.net wrote:
If my relay is in a datacenter with a fast link but constrained by a in+out<100GByte monthly limit, is it more helpful to provide:
- a slow relay that is usually available, e.g. 30KByte/sec available
about two-thirds of the time, or:
- a fast relay that hibernates a lot, e.g. 1MByte/sec but only
available 1/50th of the time?
Secondly, is it more helpful to be:
available every day for a fraction of that day, or:
available continuously for some consecutive days then hibernated for
the rest of the month?
Thanks for any advice. (I've read differing opinions on these questions written at different times, but haven't found a clear consensus.)
I don't think there is clear consensus. I would say make it available for a consecutive timespan, and set the speed limit to something not lower than 1 MB/s.
Would the sensible solution be to set RelayBandwidthBurst to a high value while leaving RelayBandwidthRate low enough to prevent hibernation? I'm not sure how this would work out, but it might result in a relay which is able to support fast connections when required and still be up most/all the time.
AVee