使用Cake 队列优化延迟

Ardentwheel commented 3 years ago

使用 Cake 队列可以优化延迟，但是对于代理，是没有用上它优先级队列功能的，既然现在开始结构优化了，也许充分利用已有的工具也是个方向大家一起测试一下看看有没有更好，希望有人测试一下bbr2+cake

优先级队列有分默认的3级和4级适配后延迟会更好，集成到路由里甚至可以进行简单的流控

diffserv3 (default)

Provides a simple, general-purpose Diffserv implementation with three tins: Bulk (CS1), 6.25% threshold, generally low priority. Best Effort (general), 100% threshold. Voice (CS7, CS6, EF, VA, TOS4), 25% threshold, reduced Codel interval.

diffserv4

Provides a general-purpose Diffserv implementation with four tins: Bulk (CS1), 6.25% threshold, generally low priority. Best Effort (general), 100% threshold. Video (AF4x, AF3x, CS3, AF2x, CS2, TOS4, TOS1), 50% threshold. Voice (CS7, CS6, EF, VA, CS5, CS4), 25% threshold.

Ubuntu 20.04 / netplan

/etc/networkd-dispatcher/routable.d/50-ifup-hooks

#!/bin/bash
# must be /bin/bash, for some reasons /bin/sh does not work

if [ "$IFACE" == "eth0" ];
then
    tc qdisc replace dev eth0 root cake ethernet
fi

查看状态 `tc -s qdisc show dev eth0`

Status

Ubuntu 20.04 Xanmod Linux 5.9 (STABLE) 2020-11-19 Vless sysctl

net.core.default_qdisc=cake
net.ipv4.tcp_congestion_control=bbr

# net.ipv4.tcp_tw_recycle=0
#net.ipv4.tcp_tw_reuse=1
#net.ipv4.ip_local_port_range = 1024 65535
#net.ipv4.tcp_rmem = 16384 262144 8388608
#net.ipv4.tcp_wmem = 32768 524288 16777216
#net.core.somaxconn = 8192
#net.core.rmem_max = 16777216
#net.core.wmem_max = 16777216
#net.core.wmem_default = 2097152
#net.ipv4.tcp_max_tw_buckets = 5000
#net.ipv4.tcp_max_syn_backlog = 10240
#net.core.netdev_max_backlog = 10240
#net.netfilter.nf_conntrack_max = 1000000
#net.ipv4.netfilter.ip_conntrack_tcp_timeout_established = 7200
#net.ipv4.tcp_slow_start_after_idle = 0
net.ipv4.tcp_notsent_lowat = 16384

`tc -s qdisc show dev eth0`

 qdisc cake 8001: root refcnt 2 bandwidth 80Mbit diffserv3 triple-isolate nonat nowash no-ack-filter split-gso rtt 200.0ms noatm overhead 38 mpu 84 
 Sent 840181829 bytes 611096 pkt (dropped 1, overlimits 296021 requeues 0) 
 backlog 0b 0p requeues 0
 memory used: 80462b of 8000000b
 capacity estimate: 80Mbit
 min/max network layer size:           32 /    1500
 min/max overhead-adjusted size:       84 /    1538
 average network hdr offset:           14

                   Bulk  Best Effort        Voice
  thresh          5Mbit       80Mbit       20Mbit
  target         10.0ms       10.0ms       10.0ms
  interval      200.0ms      200.0ms      200.0ms
  pk_delay          0us         71us         79us
  av_delay          0us          9us          8us
  sp_delay          0us          1us          2us
  backlog            0b           0b           0b
  pkts                0       610477          620
  bytes               0    840055670       127657
  way_inds            0         2190            0
  way_miss            0         3064            5
  way_cols            0            0            0
  drops               0            1            0
  marks               0            0            0
  ack_drop            0            0            0
  sp_flows            0            5            1
  bk_flows            0            1            0
  un_flows            0            0            0
  max_len             0        14980         1498
  quantum           300         1514          610