Scalability and Perf Board issueshttps://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues2020-02-07T18:18:10Zhttps://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues/14Tweak sbws scaling curves2020-02-07T18:18:10ZMike PerryTweak sbws scaling curvesAs an alternative, or perhaps in addition to, PID feedback, we could tweak the weights that sbws applies by imposing an additional curve on top of the results.
I am not a huge fan of this approach, but I am adding it for completeness.As an alternative, or perhaps in addition to, PID feedback, we could tweak the weights that sbws applies by imposing an additional curve on top of the results.
I am not a huge fan of this approach, but I am adding it for completeness.https://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues/12PID feedback for bw auths2020-02-07T18:17:32ZMike PerryPID feedback for bw authsThe Torflow PID feedback system will result in a provably optimimum load balancing equilibrium for uniform average stream capacity across the network.
The problem is if we only use bandwidth for this feedback, relays get overloaded in o...The Torflow PID feedback system will result in a provably optimimum load balancing equilibrium for uniform average stream capacity across the network.
The problem is if we only use bandwidth for this feedback, relays get overloaded in other ways. So we need to stop feedback if nodes announce that they are overloaded via memory, sockets, or CPU (which is another ticket).https://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues/11DNS test predicted circuits2020-02-07T18:15:35ZMike PerryDNS test predicted circuitsArthur and Dennis have detected that some long tail of exits/relays are really bad for unknown reasons. Sometimes this is just DNS config, but there are other things beyond that.
We could DNS test our circuits and toss out the slowest 5...Arthur and Dennis have detected that some long tail of exits/relays are really bad for unknown reasons. Sometimes this is just DNS config, but there are other things beyond that.
We could DNS test our circuits and toss out the slowest 5%, via a mechanism similar to CBT math. We don't really need to toss out a lot of circuits via this mechanism, as something like 1% or less of circuits have this problem. But the ones that do have it delay page load by 6 seconds or more.https://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues/8Peerflow/flashflow load balancing or measurement2020-02-07T18:10:42ZMike PerryPeerflow/flashflow load balancing or measurementhttps://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues/5QoS and user/circuit throttling2020-02-07T18:07:03ZMike PerryQoS and user/circuit throttlinghttps://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues/2Circuit migration (confux + reliability)2020-02-07T18:07:31ZMike PerryCircuit migration (confux + reliability)https://gitlab.torproject.org/mikeperry/scalability-and-perf-board/-/issues/1Congestion Control2020-07-09T20:14:22ZMike PerryCongestion Control