Varnish basic applications in Linux

The HTTP cache is implemented in two ways: Squid, Varnish; generally known as a cache server

Squid: Supports forward and reverse proxies, is a heavyweight cache server with very stable performance under high load conditions

Varnish: Support reverse proxy; Compared to squid is a very lightweight cache server, under high load conditions, poor performance, no squid stability; generally referred to as the HTTP accelerator;

Features of Varnish:

I. Components

Management: As a master process; provides command line interface, manages various sub-processes, initializes, loads files, etc.

Child/cache: Cache management, log data statistics, accept user requests, perform validation on all caches, clean up

Vcl:varnish configuration language; cache control mechanism, call C compiler, generate binary program, let child process load

Second, the mode of operation:

1. The requested resource can be cached:

(1) When a client requests a resource that hits the varnish cache and the cache is not expired, it responds directly to the client;

(2) Varnish as a reverse proxy cache server, when a client requests a resource that is not hit varnish cache, varnish acts as a client role to the backend host to request data, caches the requested resource first locally, and then sends the requested resource to the client.

(3) When a client requests a resource that has hit the varnish cache, but the cache on the varnish has expired, varnish will go to the backend host to confirm that the original resource was updated, and if not, the backend host responds only to the resource header to varnish. The varnish then sends the response resource to the client, and if the original resource is requested to be updated, the backend host responds to the new resource to the Varnish,varnish to cache the resource that responds locally and then responds to the client, if the requested resource backend host has been deleted. The cache in the varnish is also deleted, responding to the client 404 page

2. The requested resource is not cacheable:

(1) When the client requests the resource is not cacheable, varnish will go directly to the backend host to get the corresponding content directly to the client, instead of caching to local and then responding to the client

Measure the validity of a cache:

Cache Hit Ratio:

Document hit Ratio: Also known as a resource hit rate, measured from the number of documents

Byte hit ratio: measured from the size of the content

Cache life Cycle:

Cache entry Expiration: Lazy cleanup

Out of cache space: using the LRU algorithm (least recently using algorithms)

Note: When a cache entry expires, it is not really from the in-memory youth brigade, but is marked as invalid, not deleted, and the cache entry is actually deleted when the memory space is exhausted

Three, Cache mode:

(1) File: Cache re-disk files, self-managed critical systems

(2) malloc: Memory cache; use malloc () library call to cache a specified amount of space at varnish startup for a memory request

(3) presistent: stored on disk;

Iv. Cache freshness Detection mechanism

Validity in the validation:

If the original content changes, only the header (not the body part; The response code is 304 (not Modified)

If the original content has changed, it responds normally to the client with a response code of 200

If the original content disappears, the response is 404, and the cache entry (object) in the cache should also be deleted

Conditional Request Header:

If-modified_since: Validation based on the last modified timestamp of the original content

If-ummodified_since: If the cache has changed, the response 304

If-match: Validation based on matching of request content

If-none_match: Content verification based on ETag comparison

Five, cache control mechanism

Expiry date:

   http/1.0 Expires: The absolute length of time is given

    For example: expires:thu, June 23:38:18 GMT

http/1.1 cache-control:max-age: Given is relative duration, starting from receiving to corresponding message

Example: cache-control:max-age=700

Six, VCL-related state engine

VCL_RECV: Accept Request

Vcl_hash:

Hit:vcl_hit: Hit Cache

Miss:vcl_miss: Cache not hit

Purge:vcl_purge: Clear Cache

Pipe:vcl_pipe: pipe; direct to backend host

Pass, Hit_for_pass:vcl_pass # #跳过, request back-end host directly

Vcl_backend_fetch: Reading data

Vcl_backend_response: Back-end host response

Vcl_backend_error: Back-end host response error code

Vcl_synth: Synchronizing

Vcl_deliver: Responding to clients

Seven, the syntax format of VCL

(1) Comment://, #,/*......*/

(2) Defining subroutines

(3) Do not support loop, support condition judgment

(4) with built-in variables

(5) Use termination statement return, no return value

(6) Support operator: =, = =,! =, &&, | |

Viii. built-in variables of VCL

Req.*: Matches only requests sent by the client

req.http.*: Matches any header of the HTTP request message

BEREQ.*: HTTP request sent by varnish back-end host

Beresp.*: Header of the response message sent by the backend host to HTTP

RESP.*: Response message for HTTP response to the client by varnish

obj.*: The corresponding property of the cache stored in cache space

Common variables:

Berep:varnish back-end host request resource

Bereq.request: Request method (send request message to back-end host)

Bereq.url: Requested URL

Bereq.proto: Request Agreement

Bereq.backend: Indicates the backend host to invoke is the

BERESP: Back-end host responds to varnish resources

Beresp.proto: Response protocol

Beresp.status: Status code of the response

Beresp.reason: Response Reason

Beresp.backend.name: Backend Host name

Beresp.ttl: The remaining lifetime of the content in the backend host request

Obj.hits: The number of times this object was hit from the cache

Obj.ttl: The TTL value of the object, also known as the cache duration

Preparation Environment: CentOS 7

Varnish server	ip:172.18.42.200
Back-end Host 1: Provide HTTPD services	ip:172.18.42.201
Back-end Host 2: Provide HTTPD services	ip:172.18.42.202

One, varnish server configuration

1, installation varnish

[[email protected] ~]# yum install varnish[[email protected] ~]# ll/etc/varnish/total 20-rw-r--r--1 root root 1155 May 2 0 17:07 DEFAULT.VCL # #配置Varnish缓存策略-rw-r--r--1 root root 1224 May 11:10 default.vcl.wtc-rw-------1 root root Notoginseng Ma Y 11:09 secret-rw-r--r--1 root root (Mar) varnish.params # #配置缓存机制-rw-r--r--1 root x 20 11:1 2 Varnish.params.bak

2. Change the varnish cache mode:

[Email protected] varnish]# vim varnish.params #VARNISH_STORAGE = "file,/var/lib/varnish/varnish_storage.bin,1g" # # Default caching mechanism varnish_storage= "MALLOC,1G" # #更改缓存为malloc模式

3. Define back-end host

[Email protected] varnish]# Vim DEFAULT.VCL # #backend default {. Host = "172.18.42.202";    # #后端主机1. Port = "80"; # #web80端口}

Second, configure the back-end host

1, back-end Host 1 installation httpd, PHP

[[email protected] ~]# yum install httpd-y[[email protected] html]# systemctl start Httpd.service

2, back-end Host 2 installation httpd

[email protected] ~]# yum install nginx[[email protected] nginx]# Nginx

Third, configure the DEFAULT.VCL cache policy

1. When the request resource is cacheable, check to see if the cache is hit

Sub Vcl_deliver {if (obj.hits>0) {Set resp.http.x-cache = "hit";    } else {Set resp.http.x-cache = "MISS"; }} # #如果请求的资源未命中则显示MISS, hit will show hits

First time Request:

650) this.width=650; "src=" Http://s5.51cto.com/wyfs02/M01/80/68/wKioL1dAcE2BhLt7AABWCa7dkgA789.jpg "title=" 222. JPG "alt=" wkiol1dace2bhlt7aabwca7dkga789.jpg "/>

Second request: Hit the varnish cache

650) this.width=650; "src=" Http://s2.51cto.com/wyfs02/M01/80/6A/wKiom1dAbwqRkDH-AABdQ8JxTbo577.jpg "title=" 222. JPG "alt=" wkiom1dabwqrkdh-aabdq8jxtbo577.jpg "/>

2. Forcing the cache to be unchecked for certain resources

[Email protected] ~]# MKDIR/WEB/LWEIM/WTC # #在后端主机新目录 "WTC" [[email protected] ~]# vim/web/lweim/wtc/index.html Sub    VCL_RECV {if (req.url ~ "(? i) ^/WTC") {return (pass); }} # #当请求的url当中有 "WTC", the cache is not hit, but is directly requested by the Varnish backend host

First time Request:

650) this.width=650; "src=" http://s5.51cto.com/wyfs02/M01/80/68/wKioL1dAdXnDCyg8AABmQv8kXAU463.jpg "title=" 444. JPG "alt=" wkiol1dadxndcyg8aabmqv8kxau463.jpg "/>

Second Request:

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M00/80/6A/wKiom1dAdRei1oj3AABlZ4C8yfw079.jpg "title=" 111. JPG "alt=" wkiom1dadrei1oj3aablz4c8yfw079.jpg "/>

3. Define multiple back-end hosts

backend  web2 {    .host =  "172.18.42.202";    .port =  "80";} backend web1 {    .host =  "172.18.42.201";     . port =  "80";} sub vcl_recv {    if  (req.url ~  "(? i) \.php$")  {         set req.backend_hint = web1;   ## When PHP is available in the request URL, it is the backend host Web1 to respond                  return (pass);    }    else {         set req.backend_hint = web2;   ## Requests that do not have PHP in the URL are responded to by the backend host WEB2                  return (pass);     }} 

Request static resources for the first time

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M02/80/78/wKiom1dCrZngRHAuAABmWsflrhU854.jpg "title=" 111. JPG "alt=" wkiom1dcrzngrhauaabmwsflrhu854.jpg "/>

Second request PHP

650) this.width=650; "src=" Http://s4.51cto.com/wyfs02/M01/80/77/wKioL1dCrwTDnJSTAAClRv5XwpU328.jpg "title=" 111. JPG "alt=" wkiol1dcrwtdnjstaaclrv5xwpu328.jpg "/>

4. Achieve load Balancing

Import directors;    # #定义一个主机组backend Web1 {# #定义第一个后端服务器. Host = "172.18.42.201"; . Port = "80";}    Backend WEB2 {# #定义第二个后端服务器. Host = "172.18.42.202"; . Port = "80";}    Sub Vcl_init {# #定义一个服务器组 new Web = Directors.round_robin ();    Web.add_backend (WEB1); Web.add_backend (WEB2);}        Sub Vcl_recv {if (req.url ~ "(? i) ^/WTC") {Set req.backend_hint = Web.backend ();  return (pass); # #直接请求后端主机服务器}}

First time Request:

650) this.width=650; "src=" Http://s2.51cto.com/wyfs02/M02/80/7A/wKiom1dCta7DgpqJAABZO1KlCxQ412.jpg "title=" 555. JPG "alt=" wkiom1dcta7dgpqjaabzo1klcxq412.jpg "/>

Second Request:

650) this.width=650; "src=" http://s5.51cto.com/wyfs02/M00/80/78/wKioL1dCtr3iaNyTAABZZvVV5PY414.jpg "title=" 666. JPG "alt=" wkiol1dctr3ianytaabzzvvv5py414.jpg "/>

5, the back-end host to do health testing

Backend Web1 {. host = "172.18.42.201";    . Port = "80";        . Probe = {. url = "/";  . interval = 1s;     # #每隔多长时间检测一次. window = 8;    # #指定后端主机要检测多少次. Threshold 5;     # #指定至少要有几次是检测成功的. timeout 2s; # #指定检测时间的超时时长; default is 2s}

6. Modify Varnish run-time command parameters

[[email protected] varnish]# varnishstatmain.pools                            4         0.00           .           4.00[[email  Protected] ~]# varnishadm -s /etc/varnish/secret -t 127.0.0.1:6082param.set  thread_pools 2200      # #响应码为200即为修改成功 [[email protected]  varnish]# varnishstatmain.pools                           2          0.00          .            2.00 

7. View varnish's working status and related information

[Email protected] varnish]# Varnishstat

650) this.width=650; "src=" Http://s2.51cto.com/wyfs02/M02/80/79/wKioL1dCvQXCzSzxAAAgrd9BP8A513.png "title=" 111. PNG "alt=" Wkiol1dcvqxczszxaaagrd9bp8a513.png "/>

This article is from the "WTC" blog, so be sure to keep this source http://wangtianci.blog.51cto.com/11265133/1782214

Varnish basic applications in Linux