Summary of lvs, nginx, and haproxy transit modes

lvs, nginx, and haproxy are the three most common high-availability load balancing software. Since lvs and haproxy are not used in the current company's live network environment, although they were briefly understood and built before, they have almost forgotten it now. As for nginx load balancing, although the company is using it, it only needs to be configured once The files are all ctrl+c, ctrl+v, and I have almost forgotten the theoretical things about forwarding. The only words that vaguely come to mind now are upstream, dr, and ip_hash. Now we will summarize the forwarding methods of the three.

1. LVS forwarding mode

LVS is a high-possibility software that works on the fourth layer written by Dr. Zhang Wensong. Unlike the latter two, it does not support seven-layer forwarding, but precisely because of its simplicity, it is the most stable. There are three IP load balancing technologies: VS/NAT (Virtual Server via Network Address Translation), VS/TUN (Virtual Server via IP Tunneling) and VS/DR (Direct Routing). The specific comparison between the three is shown in the table below.

2. nginx load mode

nginx has five load algorithm modes, namely: polling, weight (weight), ip_hash, fair, url_hash. Now explain one by one:

1. Polling (default): Each request is assigned to different back-end servers one by one in chronological order. If the back-end server goes down, it can be automatically eliminated.
2. weight: Specify the polling probability. The weight is proportional to the access ratio. It is used when the back-end server performance is uneven. Configured as:
   1. upstream bakend {
   2. server 192.168.0.14 weight=10;
   3. server 192.168.0.15 weight=10;
   4. }
3. ip_hash: Each request is allocated according to the hash result of the access IP, so that each visitor has fixed access to a back-end server, which can solve the session problem. Configured as:
   1. upstream bakend {
   2. ip_hash;
   3. server 192.168.0.14: 88;
   4. server 192.168. 0.15:80;
   5. }
4. fair: Allocate requests according to the response time of the backend server, and those with short response times will be allocated first.
   1. upstream backend {
   2. server server1;
   3. server server2;
   4. fair;
   5. }
5. url_hash: by visiting url The hash result is used to distribute requests so that each URL is directed to the same backend server. It is more effective when the backend server is cached. Configuration like:
   1. upstream backend {
   2. server squid1:3128;
   3. server squid2:3128 ;
   4. hash $request_uri;
   5. hash_method crc32;
   6. }

Note: In the fifth mode, you need to pay attention to adding hash statements in upstream. Other parameters such as weight cannot be written in the server statement. hash_method is used hash algorithm. The parameters often followed by

server are as follows:

• down means that the previous server will not participate in the load temporarily
• weight The default is 1. The larger the weight, the greater the weight of the load.
• max_fails: The number of allowed request failures defaults to 1. When the maximum number is exceeded, the error defined by the proxy_next_upstream module is returned
• fail_timeout: The pause time after max_fails failures.
• backup: When all other non-backup machines are down or busy, request the backup machine. So this machine will have the least pressure.

Three, haproxy

haproxy has the most loaded algorithms among the three, with eight types, so it has the most application scenarios and the most flexible configuration. The specific eight algorithms are:

①roundrobin, which means simple Polling, this is basically what load balancing has;

②static-rr, which means based on weight, is similar to nginx's weight algorithm;

③leastconn, which means the least connected person is processed first, a bit similar to nginx's fair, but fair It is based on the response time;

④source, which means based on the request source IP. This is similar to the IP_hash mechanism of Nginx. We use it as a method to solve the session problem. It is recommended to pay attention;

⑤ri, which means based on the URI of the request, similar to nginx's url_hash;

⑥rl_param, which means that 'balance url_param' requires an URL parameter name according to the requested URL parameter;

⑦hdr(name), which means that each HTTP request is locked based on the HTTP request header;

⑧rdp-cookie(name) means locking and hashing each TCP request based on cookie(name).

IV. Summary

Specific implementation network applications can choose the best load method based on the actual situation of the object. Among the three, lvs has the best stability and the least configurability; nginx is the strongest in regular matching of domain names and directory structures. At the same time, it has less dependence on the network, but its performance is slightly worse than lvs and haproxy. ; haproxy supports virtual hosts, especially in terms of session retention. It has three algorithms to achieve session sharing - IP identification (source), cookie identification, and session identification. In addition, it does mysql This software is also often used in HA solutions.

The above has introduced a summary of the transfer modes of lvs, nginx, and haproxy, including relevant content. I hope it will be helpful to friends who are interested in PHP tutorials.