Linux系统网络性能优化与负载均衡指南

一、网络性能优化

1. 内核参数调优

# 编辑/etc/sysctl.conf文件
# 增加或修改以下参数

# 增大TCP窗口大小
net.core.rmem_max = 16777216
net.core.wmem_max = 16777216
net.ipv4.tcp_rmem = 4096 87380 16777216
net.ipv4.tcp_wmem = 4096 65536 16777216

# 启用TCP窗口缩放
net.ipv4.tcp_window_scaling = 1

# 优化TIME_WAIT状态
net.ipv4.tcp_fin_timeout = 30
net.ipv4.tcp_tw_reuse = 1
net.ipv4.tcp_tw_recycle = 1

# 增大连接跟踪表大小
net.ipv4.ip_conntrack_max = 65536
net.ipv4.netfilter.ip_conntrack_max = 65536

# 禁用ICMP重定向
net.ipv4.conf.all.accept_redirects = 0
net.ipv4.conf.all.send_redirects = 0

# 应用修改
sysctl -p

2. 中断亲和性设置

# 查看网络接口的中断号
cat /proc/interrupts | grep eth

# 设置CPU亲和性(假设使用CPU0处理eth0中断)
echo 1 > /proc/irq/[中断号]/smp_affinity

3. 网络接口调优

# 启用巨帧(如果网络支持)
ifconfig eth0 mtu 9000

# 调整队列长度
ifconfig eth0 txqueuelen 10000

# 启用多队列RSS(需要硬件支持)
ethtool -L eth0 combined 8

4. TCP协议栈优化

# 启用TCP快速打开
echo 3 > /proc/sys/net/ipv4/tcp_fastopen

# 调整拥塞控制算法(可选cubic, reno, bbr等)
echo "bbr" > /proc/sys/net/ipv4/tcp_congestion_control

# 增大最大半连接队列
echo 4096 > /proc/sys/net/ipv4/tcp_max_syn_backlog

二、负载均衡实现

1. 基于IPVS的负载均衡

# 安装IPVS
apt-get install ipvsadm  # Debian/Ubuntu
yum install ipvsadm      # CentOS/RHEL

# 配置负载均衡规则
ipvsadm -A -t 192.168.1.100:80 -s rr
ipvsadm -a -t 192.168.1.100:80 -r 192.168.1.101:80 -g
ipvsadm -a -t 192.168.1.100:80 -r 192.168.1.102:80 -g

# 查看状态
ipvsadm -ln

2. 基于Nginx的负载均衡

http {
    upstream backend {
        server 192.168.1.101:80 weight=5;
        server 192.168.1.102:80;
        server 192.168.1.103:80 backup;

        # 负载均衡算法
        # least_conn;   # 最少连接
        # ip_hash;      # IP哈希
    }

    server {
        listen 80;

        location / {
            proxy_pass http://backend;
            proxy_set_header Host $host;
            proxy_set_header X-Real-IP $remote_addr;
        }
    }
}

3. 基于HAProxy的负载均衡

frontend http-in
    bind *:80
    default_backend servers

backend servers
    balance roundrobin
    server server1 192.168.1.101:80 check
    server server2 192.168.1.102:80 check
    server server3 192.168.1.103:80 check backup

    # 健康检查
    option httpchk GET /health
    http-check expect status 200

4. 基于Keepalived的高可用

# 安装Keepalived
apt-get install keepalived  # Debian/Ubuntu
yum install keepalived      # CentOS/RHEL

# 配置文件(/etc/keepalived/keepalived.conf)
vrrp_instance VI_1 {
    state MASTER            # 另一台设为BACKUP
    interface eth0
    virtual_router_id 51
    priority 100            # BACKUP设为较低值如90
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
    }
    virtual_ipaddress {
        192.168.1.100/24
    }
}

三、监控与维护

1. 网络性能监控工具

# 实时监控网络流量
iftop -i eth0

# 查看网络连接状态
ss -s
netstat -s

# 带宽测试
iperf3 -s  # 服务器端
iperf3 -c server_ip  # 客户端

2. 负载均衡监控

# IPVS状态监控
ipvsadm -ln --stats

# Nginx状态(需配置status模块)
location /nginx_status {
    stub_status on;
    access_log off;
    allow 127.0.0.1;
    deny all;
}

# HAProxy状态
listen stats
    bind *:1936
    stats enable
    stats uri /
    stats hide-version
    stats auth admin:password

四、高级优化技术

1. DPDK加速

# 安装DPDK
git clone http://dpdk.org/git/dpdk
cd dpdk
make config T=x86_64-native-linuxapp-gcc
make -j

# 绑定网卡到DPDK
./usertools/dpdk-devbind.py --bind=vfio-pci eth1

2. XDP(eXpress Data Path)

# 编译XDP程序
clang -O2 -target bpf -c xdp_prog.c -o xdp_prog.o

# 加载到网卡
ip link set dev eth0 xdp obj xdp_prog.o sec xdp

3. 多路径TCP(MPTCP)

# 启用MPTCP
echo 1 > /proc/sys/net/mptcp/enabled

# 配置路径管理器
echo "fullmesh" > /proc/sys/net/mptcp/mptcp_path_manager

通过以上优化和配置，可以显著提升Linux系统的网络性能和实现高效的负载均衡。具体实施时应根据实际网络环境和应用需求进行调整。