Re: GD to database directly

Eric Butera wrote:

On 7/11/06, Kevin Waterson <kevin@oceania.net> wrote:

How much of a performance hit? Kevin -- "Democracy is two wolves and a lamb voting on what to have for lunch. Liberty is a well-armed lamb contesting the vote." -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php

Here's to being bored on lunch. :) Now to add my two cents to the convo: I don't see a problem with storing images in the DB IF they aren't going to be continually accessed that way. For example say you have a script that lets a user upload an image and creates a small, medium, and large view out of it. Stick the original in the Db and then make your 3 sizes on the fly and cache the output so they are not continually created. This way you could down the road re-do all the sizes if you go through a site redesign or the client chooses they want them 450 instead of 400. =) Via MySQL 4.1 & PHP 5.1.4: erics:~ eric$ ab -n500 -c5 http://localhost/fs_vs_db/viewimage_db.php This is ApacheBench, Version 1.3d <$Revision: 1.73 $> apache-1.3 Copyright (c) 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/ Copyright (c) 1998-2002 The Apache Software Foundation, http://www.apache.org/ Benchmarking localhost (be patient) Completed 100 requests Completed 200 requests Completed 300 requests Completed 400 requests Finished 500 requests

Server Software:        Apache/1.3.33
Server Hostname:        localhost
Server Port:            80

Document Path:          /fs_vs_db/viewimage_db.php
Document Length:        25256 bytes

Concurrency Level:      5

Time taken for tests: 4.323 seconds

Complete requests:      500
Failed requests:        0
Broken pipe errors:     0
Total transferred:      12727000 bytes
HTML transferred:       12628000 bytes
Requests per second:    115.66 [#/sec] (mean)
Time per request:       43.23 [ms] (mean)
Time per request:       8.65 [ms] (mean, across all concurrent requests)
Transfer rate:          2944.02 [Kbytes/sec] received

Connnection Times (ms)

             min  mean[+/-sd] median   max
Connect:        0     0    0.4      0    10
Processing:    14    42   12.6     38   147
Waiting:       14    42   12.6     38   147
Total:         14    42   12.6     38   147

Percentage of the requests served within a certain time (ms)

 50%     38
 66%     40
 75%     44
 80%     46
 90%     56
 95%     67
 98%     78
 99%     97
100%    147 (last request)

Raw Image: erics:~ eric$ ab -n500 -c5 http://localhost/fs_vs_db/cloud.jpg This is ApacheBench, Version 1.3d <$Revision: 1.73 $> apache-1.3 Copyright (c) 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/ Copyright (c) 1998-2002 The Apache Software Foundation, http://www.apache.org/ Benchmarking localhost (be patient) Completed 100 requests Completed 200 requests Completed 300 requests Completed 400 requests Finished 500 requests

Server Software:        Apache/1.3.33
Server Hostname:        localhost
Server Port:            80

Document Path:          /fs_vs_db/cloud.jpg
Document Length:        25256 bytes

Concurrency Level:      5

Time taken for tests: 0.624 seconds

Complete requests:      500
Failed requests:        0
Broken pipe errors:     0
Total transferred:      12775000 bytes
HTML transferred:       12628000 bytes
Requests per second:    801.28 [#/sec] (mean)
Time per request:       6.24 [ms] (mean)
Time per request:       1.25 [ms] (mean, across all concurrent requests)
Transfer rate:          20472.76 [Kbytes/sec] received

Connnection Times (ms)

             min  mean[+/-sd] median   max
Connect:        0     1    0.7      1     5
Processing:     3     5    6.6      4   131
Waiting:        1     4    6.7      3   130
Total:          3     6    6.5      5   131

Percentage of the requests served within a certain time (ms)

 50%      5
 66%      5
 75%      5
 80%      6
 90%      7
 95%     10
 98%     11
 99%     19
100%    131 (last request)

The time taken per request, though (and that's about all we can get with a concurrency as low as 5) doesn't tell us much. We also don't know exactly what the PHP code is doing, how it does it, how your database is organized/indexed/accessed, if you have any PHP accelerators installed (Normally the PHP script would be reconverted to bytecode every execution), etc. Additionally, your test isn't really MySQL versus filesystem, it's PHP+MySQL versus filesystem. Perhaps a more useful comparison would be PHP+Filesystem versus PHP+MySQL. As in, the same PHP script for both benchmarks, except one copy uses file_get_contents and echo (Closer match than readfile, since MySQL would require loading the file into memory) and the other uses MySQL. This would be a closer match that would tell us how much latency is induced by the actual database itself rather than PHP and loading stuff into memory. Regards, Adam.