Varnish acceleration website Image Display

Since the dynamic generation of thumbnails went live, NFS client logs often reported an error that could not be responded. I suspect that the NFS machine Io is too large, so I wanted to use varnish to decompress the NFS machine and store the thumbnails in the memory, reduce the read pressure, and then there will be a blog post. Summary: varnish is a high-performance open-source HTTP accelerator. It is rated more powerful than squid on the Internet, and the installation and configuration are fast and convenient.
1. installation:

# Wget http://repo.varnish-cache.org/source/varnish-2.1.5.tar.gz
# Tar zxvf varnish-2.1.5.tar.gz CD Varnish-2.1.5
#./Configure-Prefix =/usr/local/Varnish
# Make & make install

Ii. configuration file:

# Cp/usr/local/Varnish/etc/Varnish/Default. VCL/usr/local/Varnish/etc/Varnish/Default. VCL. Bak
# Vi/usr/local/Varnish/etc/Varnish/Default. VCL

# Backend Web Server IP address and port of the reverse proxy request
Backend default {
. Host = "192.168.1.1 ";
. Port = "80 ";
}

Sub vcl_recv {
If (req. Request! = "Get" & Req. Request! = "Head "){
Return (PIPE );
} Elseif (req. url ~ "\. (JPG | GIF | PNG) $ "){
Return (lookup );
} Else {
Return (PASS );
}
}

Sub vcl_pipe {
Return (PIPE );
}

Sub vcl_pass {
Return (PASS );
}

Sub vcl_hash {
Set Req. Hash + = Req. url;
If (req. http. HOST ){
Set Req. Hash + = Req. http. Host;
} Else {
Set Req. Hash + = server. IP;
}
Return (hash );
}

Sub vcl_hit {
If (! OBJ. cacheable ){
Return (PASS );
}
Return (deliver );
}

Sub vcl_miss {
Return (FETCH );
}

Sub vcl_fetch {
If (beresp. http. Pragma ~ "No-Cache" |
Beresp. http. cache-control ~ "No-Cache" |
Beresp. http. cache-control ~ "Private "){
Return (PASS );
}
# Clear the cookie and set the expiration time to 30 days
If (req. Request = "get" & Req. url ~ "\. (JPG | GIF | PNG) $ "){
Unset Req. http. Cookie;
Set beresp. TTL = 30d;
}
Return (deliver );
}

Sub vcl_deliver {
# Add an identifier (which can be viewed using firebug) x-hits to indicate whether it is hit
Set resp. http. X-hits = obj. Hits;
If (obj. Hits> 0 ){
Set resp. http. X-Cache = "H ";
} Else {
Set resp. http. X-Cache = "M ";
}
}

Iii. startup:

#/Usr/local/Varnish/sbin/varnishd-F/usr/local/Varnish/etc/Varnish/default. VCL-N/var/log/Varnish-s malloc, 15g-W 2000, 51200, 20-T 127.0.0.1: 8080-A 0.0.0.0:

Note:
-F configuration file path
-N main directory
-S allocates 15 GB of memory
-Minimum and Maximum number of W threads
-T manages IP addresses and ports. You can use Telnet 127.0.0.1 2000 to manage varnish.
-A listens to port 8080

4. Write log files:

#/Usr/local/Varnish/bin/varnishncsa-N/var/log/Varnish-W/var/log/Varnish/varnish. Log

5. View zookeeper:

#/Usr/local/Varnish/bin/varnishstat-N/var/log/Varnish

Vi. About varnish:

# Pkill varnishd

7. Clear the inventory:

1. Clear the specified domain name:

/Usr/local/Varnish/bin/varnishadm-T 127.0.0.1: 2000 purge "Req. http. Host ~ Statics.xxx.com.tw"

2. Clear all the caches in the image directory:

/Usr/local/Varnish/bin/varnishadm-T 127.0.0.1: 2000 URL. Purge/image/

3. Clear all caches:

/Usr/local/Varnish/bin/varnishadm-T 127.0.0.1: 2000 URL. Purge * $

8. test:
AB test:
AB-C 100-N 1000 http://x.x.x.x/129974432673506402_128x92.jpg
Before varnish is used:

Server Software: nginx
Server Hostname: x. x
Server port: 8000

Document path: 129974432673506402_128x92.jpg
Document length: 2443 bytes

Concurrency level: 100
Time taken for tests: 0.667390 seconds
Complete requests: 1000
Failed requests: 0
Write errors: 0
Total transferred: 2730000 bytes
HTML transferred: 2443000 bytes
Requests per secondd: 1498.37 [#/sec] (mean)
Time per request: 66.739 [MS] (mean)
Time per request: 0.667 [MS] (mean, internal SS all concurrent requests)
Transfer Rate: 3994.67 [Kbytes/sec] canceled ed

Connection times (MS)
Min mean [+/-SD] median Max
Connect: 0 0 0.6 0 4
Processing: 2 61 23.2 57 248
Waiting: 1 60 23.2 57 247
Total: 2 61 23.1 58 248

Percentage of the requests served within a certain time (MS)
50% 58
66% 65
75% 73
80% 77
90% 89
95% 100
98% 116
99% 139
100% 248 (longest request)

After varnish is used:

Server Software: nginx
Server Hostname: p2.591.com.tw
Server port: 8000

Document path: 129974432673506402_128x92.jpg
Document length: 2443 bytes

Concurrency level: 100
Time taken for tests: 0.271769 seconds
Complete requests: 1000
Failed requests: 0
Write errors: 0
Total transferred: 2795675 bytes
HTML transferred: 2445443 bytes
Requests per secondd: 3679.60 [#/sec] (mean)
Time per request: 27.177 [MS] (mean)
Time per request: 0.272 [MS] (mean, internal SS all concurrent requests)
Transfer Rate: 10045.30 [Kbytes/sec] canceled ed

Connection times (MS)
Min mean [+/-SD] median Max
Connect: 0 6 2.3 6 12
Processing: 1 7 2.3 7 17
Waiting: 1 6 2.1 7 15
Total: 3 13 1.7 14 23

Percentage of the requests served within a certain time (MS)
50% 14
66% 14
75% 14
80% 14
90% 15
95% 15
98% 16
99% 17
100% 23 (longest request)

Summary:
After varnish went online, after observing for a while, we found that the NFS error cause has not been improved. Now we have switched the generated image part to NFS, and fewer NFS errors are reported. Next, switch to a Gigabit Switch.

Exam:
Varnish cache + Riak
Http://blog.dloh.org/2011/01/varnish-cache-riak.html)
Installation and configuration of varnish cache reverse proxy server and HTTP accelerator
Http://blog.51yip.com/cache/618.html)
Varnish hands
Http://www.varnish-cache.org/docs/2.1/tutorial/vcl.html)
Varnish installation process details
(Http://thinkgroupon.com/redirect.php? Tid = 3872 & goto = lastpost)
Use varnish to replace squid for detailed solution of website cache accelerator [original]
Http://blog.s135.com/post/313)
Summary of varnish usage
Http://iyubo.blogbus.com/logs/35085709.html)