Dolphin benchmark on php 7

So i ran some benchmarks of dolphin on php7 and this is the result.

 

i used this to test

ab -k -c 10 -n 2000 http://testsite.example

 

on php 5.5.9

Time taken for tests: 47.495 seconds
Requests per second: 42.11 [#/sec] (mean)
Time per request: 237.476 [ms] (mean)
Time per request: 23.748 [ms] (mean, across all concurrent requests)

on php 7.0.4

Time taken for tests:   25.804 seconds
Requests per second:    77.51 [#/sec] (mean)
Time per request:       129.020 [ms] (mean)
Time per request:       12.902 [ms] (mean, across all concurrent requests)

 

So thats about 1.8x speed improvement on a fresh install without doing any modification or module install. This gap will "probably" increase as more modules are installed.

 

Just thought i would share for anyone interested.

Peace!

so much to do....
Quote · 7 Mar 2016

Activate zend opcache and make a new test to see with how much you will improve this test.

Quote · 7 Mar 2016

 

Activate zend opcache and make a new test to see with how much you will improve this test.

 Its enabled by default.

so much to do....
Quote · 7 Mar 2016

It would be nice to see what the test results are without an opcode cache.

Geeks, making the world a better place
Quote · 7 Mar 2016

some people said dolphin won't work on php 7. great news Smile

Always remember that the future comes one day at a time.
Quote · 7 Mar 2016

 

some people said dolphin won't work on php 7. great news Smile

I was one of them. It's actually a dev build, sorry should have mentioned that above.

https://github.com/boonex/dolphin.pro/pull/362

so much to do....
Quote · 8 Mar 2016

 

It would be nice to see what the test results are without an opcode cache.

 I guess but too late for that, I scraped the server. I will try that later maybe but it's kind of a pita to compile mysql for php7 to get it to work.

so much to do....
Quote · 8 Mar 2016

 

 

It would be nice to see what the test results are without an opcode cache.

 I guess but too late for that, I scraped the server. I will try that later maybe but it's kind of a pita to compile mysql for php7 to get it to work.

Have you looked at MariaDB?  I am using it for my Dolphin and not seeing any issues; it is a drop in replacement for MySQL and offers benefits over MySQL; The developers have moved away from trying to have the same numerical release as MySQL because they could move ahead with implementing features. The same php commands work with MariaDB so no need to change any code.

Geeks, making the world a better place
Quote · 8 Mar 2016

 

 

 

It would be nice to see what the test results are without an opcode cache.

 I guess but too late for that, I scraped the server. I will try that later maybe but it's kind of a pita to compile mysql for php7 to get it to work.

Have you looked at MariaDB?  I am using it for my Dolphin and not seeing any issues; it is a drop in replacement for MySQL and offers benefits over MySQL; The developers have moved away from trying to have the same numerical release as MySQL because they could move ahead with implementing features. The same php commands work with MariaDB so no need to change any code.

 When I said MySQL I meant php MySQL extension (mysql_* functions), not the server or client.

so much to do....
Quote · 8 Mar 2016

Hi,

 

Which version of dolphin was used for this php 7 test ?

 

Because most of forum posts says php 7 is not compatible with dolphin because of the mysql_connect (stuff removed in php 7...)

Quote · 27 Mar 2016

i used dolphin version 7.3 with php 7 - and after making a lot of changes of mysql to mysqli - as well as other corresponding coding - i found it is a bit more than just changingthat and realized it would require a rewrite of a lot of dolphins core functions - too much work.

caredesign.net
Quote · 27 Mar 2016

 

Hi,

 

Which version of dolphin was used for this php 7 test ?

 

Because most of forum posts says php 7 is not compatible with dolphin because of the mysql_connect (stuff removed in php 7...)

 It's a dev build ^ for 7.4/leap release.

so much to do....
Quote · 27 Mar 2016

Revisiting this as I build a new server.  I see that php 6 was removed because of lots of issues; instead they jumped to php 7.  From what I see, they are removing the deprecated items that will eventually need to be changed in Dolphin.  However, they are still supporting the 5.x source with the current 5.6.20. so users of Dolphin don't have any worries yet.  My decision is to put 5.6.20 on the server.

Geeks, making the world a better place
Quote · 27 Apr 2016

 Hey Prashank, I meant to ask whether that was on a normal hard drive (7200rpm) or SSD?

Im guessing normal HDD right?

So i ran some benchmarks of dolphin on php7 and this is the result.

 

on php 5.5.9

Time taken for tests: 47.495 seconds
Requests per second: 42.11 [#/sec] (mean)
Time per request: 237.476 [ms] (mean)
Time per request: 23.748 [ms] (mean, across all concurrent requests)

on php 7.0.4

Time taken for tests:   25.804 seconds
Requests per second:    77.51 [#/sec] (mean)
Time per request:       129.020 [ms] (mean)
Time per request:       12.902 [ms] (mean, across all concurrent requests)

 

So thats about 1.8x speed improvement on a fresh install without doing any modification or module install. This gap will "probably" increase as more modules are installed.

 

 

Ultra Newb reporting for duty.
Quote · 27 Apr 2016

It was on a $5 VM from https://www.digitalocean.com/ which is SSD. I love SSDs.

so much to do....
Quote · 28 Apr 2016

SSD may mean it gets off the server's hard disc faster but the bottleneck of web delivery does not change with SSD.

Geeks, making the world a better place
Quote · 28 Apr 2016

 

SSD may mean it gets off the server's hard disc faster but the bottleneck of web delivery does not change with SSD.

 Curious, what would that bottleneck be? Undecided

so much to do....
Quote · 29 Apr 2016

Well, let's say I am on a dialup with a 56K modem; do you think the webpage would load quicker for me if a SSD was used over a platter based hard disc; I doubt I would see much if any difference.  If I am watching a video, that video has to be broken up into packets and sent to me and it won't happen any faster than what my connection to your website is going to allow.

 

If you are building your own servers and have your own connection to the backbone, then go for it.  However, until these rental hosting companies offer 3TB SSD for the same price as platter based drives; it is just not worth the extra costs and if you are planning on having a video heavy site, you will quickly fill up those tiny SSDs that is currently being offered by many rental companies.

Geeks, making the world a better place
Quote · 29 Apr 2016

 

Well, let's say I am on a dialup with a 56K modem; do you think the webpage would load quicker for me if a SSD was used over a platter based hard disc; I doubt I would see much if any difference.  If I am watching a video, that video has to be broken up into packets and sent to me and it won't happen any faster than what my connection to your website is going to allow.

 

If you are building your own servers and have your own connection to the backbone, then go for it.  However, until these rental hosting companies offer 3TB SSD for the same price as platter based drives; it is just not worth the extra costs and if you are planning on having a video heavy site, you will quickly fill up those tiny SSDs that is currently being offered by many rental companies.

Haha, i kinda knew you were gonna say it was the network, although its true that network speed is almost always the bottleneck but thats not something anyone can optimize for if someone is still using 56k, its their problem really.

 

When serving large files the access time doesn't matter, so a HDD is fine. Its the random read/writes where SSDs destroys a HDD before it even notices. In cases where your database doesn't fit in memory or not is configured to do so (should always try to fit the whole database in memory for blazing fast reads, if you're trying to speed up a database, forget SSD and buy more friggin memory as many as you can install on the mobo :p)

OR

suppose when your server is getting hit with a large number of traffic and you're serving lots of little images (like a bazillion emoticons) or something like that.

 

that is when you will start to notice difference of what ~100 random IOPS vs ~100,000 random IOPS actually do and yes, you will notice the difference even on a 56k.

 

We're off-topic and I was bored, so there we go :/

 

Also, another question, have you tried SSD on a desktop or server before?

so much to do....
Quote · 29 Apr 2016

Where can I download the 7.4 beta ?

I'M interested to test it

 

 

Hi,

 

Which version of dolphin was used for this php 7 test ?

 

Because most of forum posts says php 7 is not compatible with dolphin because of the mysql_connect (stuff removed in php 7...)

 It's a dev build ^ for 7.4/leap release.

 

Quote · 7 Sep 2016

 

Where can I download the 7.4 beta ?

The new beta is v.10, you can download it here:

https://github.com/boonex/dolphin.pro/archive/master.zip 

Rules → http://www.boonex.com/terms
Quote · 10 Sep 2016

 Thanks ;)

 

 

Where can I download the 7.4 beta ?

The new beta is v.10, you can download it here:

https://github.com/boonex/dolphin.pro/archive/master.zip 

 

Quote · 11 Sep 2016

Hi,

 

In https://github.com/boonex/dolphin.pro/archive/master.zip  in the folder inc/ in version.inc.php it says $site['ver'] = '7.3';
$site['build'] = '0';

 

Is it the 7.3 or 7.4 dev ?

 

I installed this version in my server that have php 7.0.10 and it's is working well and FAST ;)

Quote · 24 Sep 2016
 
 
Below is the legacy version of the Boonex site, maintained for Dolphin.Pro 7.x support.
The new Dolphin solution is powered by UNA Community Management System.