Jump to content

Prestashop 1.5.0.9 speed compared to 1.4.7


Recommended Posts

Hi

I have just started testing a version of 1.5 as a final release seems to be getting closer and configured a very basic shop with the demo products and was amazed at how slow it appears compared with my current 1.4.7 installation. This was all done on a local test machine but should give a good comparison.

 

Without going into the many additional ways of improving speed beyond those easily achieved in the BO UI, has anyone else noticed the speed decrease in 1.5 and does anyone know if anything will be done before final release.

 

Thanks

Link to comment
Share on other sites

Hi

I find both a lot slower, even after the BO pages have loaded for the first time. It may be faster when in a production environment but my initial view is that it is unuseable compared to 1.4 versions.

 

I aim to do further comparisons but this may not be for a few weeks due to other commitments.

 

Paul

Link to comment
Share on other sites

You can get an idea of the performance difference by comparing the loading times from 1.4 and 1.5 that appear at the bottom left of each page (BO only). You can get more data by activating _PS_DEBUG_PROFILING_ in file /config/defines.inc.php, or with network analysis tools like the network tab of firebug.

Link to comment
Share on other sites

Thanks Thomas! But that isn't quite scientific and would probably differ for everyone as well.

 

What I meant was, could you include that aspect as part of your testing suite so that over time you (and we) can look at objective speed benchmarks of different versions.

Link to comment
Share on other sites

Hello,

 

I am developping a website using 1.5. The speed (frontend) is very slow (around 5 sec then 2 sec actualizing) and I was wondering if the only reason was that I did'nt have a dedicated server.

 

But no : to be sure I installed 1.5 and 1.4 with the same parameters (smarty -> no compilation ; cache activated), and there is a real speed difference.

 

1.5 was 5 sec 1st time then 2 sec actualizing

1.4 was 2,5 sec 1st time then 0,7 sec actualizing !!

 

Please note these speeds are specific only to my server, but I really wonder if speed will be improved for future realeses.

 

Do we have any update from Prestashop on this topic ?

 

Many thanks

Link to comment
Share on other sites

Hello,

 

I am developping a website using 1.5. The speed (frontend) is very slow (around 5 sec then 2 sec actualizing) and I was wondering if the only reason was that I did'nt have a dedicated server.

 

But no : to be sure I installed 1.5 and 1.4 with the same parameters (smarty -> no compilation ; cache activated), and there is a real speed difference.

 

1.5 was 5 sec 1st time then 2 sec actualizing

1.4 was 2,5 sec 1st time then 0,7 sec actualizing !!

 

Please note these speeds are specific only to my server, but I really wonder if speed will be improved for future realeses.

 

Do we have any update from Prestashop on this topic ?

 

Many thanks

 

I don't represent prestashop. But i set up a test site with the SVN version of 1.5 and subjectively i don't have any problems with the speed.

 

http://184.173.230.160/~prestate/index.php

 

I honestly don't know what kind of hosting people who are complaining about the speed are using. I'm just using regular shared hosting.

Link to comment
Share on other sites

  • 2 weeks later...

Hi

The type of hosting being used is not really a factor when doing a straight comparison between 1.4 and 1.5 on the same host, whether local or in a production environment. Using a local environment and identical configurations of 1.4.6.2 and 1.5.0.9, the 1.5 version BO seems to be about 2 times slower, sometimes 3 times slower after the first page load to create the Smarty cache. This is based solely on the timings shown on the screen.

 

I will try and do more testing of the FO.

Link to comment
Share on other sites

Hi ponddude,

I am also testing 1.5. There are scenarios when it becomes horrible slow. Afaik missing pictures or mysql errors. When debug

mode is off you do not see these errors. For the clean install with demo products I can not complain that much about speed but when there are mysql errors front and backend are gettting extremely slow. Anyway I guess that any cache works that way you are complaining about ... in reality search bots, customers tec.pp. are constantly hitting you website so I do not think something like a cache preloader is necessary to boost performance. Anyway P.S. 1.5 is still beta and so maybe to early to judge speed at the moment. Regards, trip

Link to comment
Share on other sites

To cube-in,

I guess your image paths or whatever are wrong.

That is the reason why site loads so slow but in case in misunderstood you... regenerating 18000 pictures is

not a problem of Prestashop... try it with Photoshop and it will take a long time anyway.... ich you want to do it

with prestashop then you have to increase php script execution time.. maybe 1 or more hours.

or use batch processing from irfanview or other freeware.

Regards, trip

Link to comment
Share on other sites

I made the painfull observation, that all latest versions from 1.4.6.2 onwards are slow. I've done all possible improvements and optimizations available and the parse is not lower than 2 seconds on products/images. Today I downgraded my shop (19.000 products, 400 categories, 200 filter optioms) to 1.4.5.1. This is the latest version working with TTFB under 1 second. All other versiones are slow and slower. Including 1.5.

  • Like 1
Link to comment
Share on other sites

Hi all

 

This is my first post in this forum, having only in the last few days installed both 1.4.7 and 1.5.0.9 onto the same host with the same instance of PHP, Apache2 and MySQL.

 

Using Firefox network diagnostics, I can confirm that 1.5 is indeed taking twice if not more time to present a page (using same demo data) and this is confirmed with both BO's.

 

Having looked at TroyRobs site I can confirm that it is very quick compared to my own 1.5 site

 

My sites are on a single VMWare install of Ubuntu 12.4 with 1GB RAM, 'top' shows very light use.

 

What I would be very interested in is if the more experienced members of the forum could advise a methodology for identifying any bottlenecks in the PS/PHP/Apache2/Mysql chain so that any errors can be identified

Link to comment
Share on other sites

The default installation with sample categories and sample products is not a problem. The problem is when you have a catalog with more thant 2.000 or at least 5.000 products. Than you will really notice the speed problem. And it is true that since version 1.4.6.2 the speed is a problem for all versions onwards. I compared an empty 1.4.4.1 on the same server, all the same technical equipment and versions after 1.4.6.2. Double of TTFB and also speed loss double. With 1.5. the same problems, double/triple loss by opening products with 3 or more pictures and some product attributes.

 

I'm on a fast-cgi vHost with 256 MB RAM and the max. of balance load a single host can have on a network. So it's not a server issue. Perhaps the page of troybord is yet alone on the webspace, without other neighbours and sites with PR 5 and 6... By trying the page on one of the products I had a peak from 3 seconds. It was just a peak on a shop with some categories, some products and some pictures. You cannot compare this with running shops and big database, filteroptions and high quantity of products.

Link to comment
Share on other sites

The default installation with sample categories and sample products is not a problem. The problem is when you have a catalog with more thant 2.000 or at least 5.000 products. Than you will really notice the speed problem. And it is true that since version 1.4.6.2 the speed is a problem for all versions onwards. I compared an empty 1.4.4.1 on the same server, all the same technical equipment and versions after 1.4.6.2. Double of TTFB and also speed loss double. With 1.5. the same problems, double/triple loss by opening products with 3 or more pictures and some product attributes.

 

I'm on a fast-cgi vHost with 256 MB RAM and the max. of balance load a single host can have on a network. So it's not a server issue. Perhaps the page of troybord is yet alone on the webspace, without other neighbours and sites with PR 5 and 6... By trying the page on one of the products I had a peak from 3 seconds. It was just a peak on a shop with some categories, some products and some pictures. You cannot compare this with running shops and big database, filteroptions and high quantity of products.

 

is it because the mysql type difference MyISAM InnoDB?

Link to comment
Share on other sites

Many things affect the speed. Do not try to compare it only once to obtain accurate results. in my tests results for PS 1.5 beta version is not too much difference in terms of speed of loading, but this's something normally that still need to load more when compared with ps 1.4. It is therefore that now still in beta :P . Hope so this issue in stable 1.5 version can be optimized

Link to comment
Share on other sites

Hi Prestashop Mania, unfortunately the links you provided will not work

 

But getting back to the tests, my comparison was a straightforward A:B test and I would like to have a better understanding of why the speed difference. The customer or end user experience is of course the ultimate test and I would not want to devote a great deal of time understanding PS1.5 if my customers end up not using my website because of delays in presenting pages

 

I did ask the question as to why TroyRob’s remote site can be more responsive than my local site given that TroyRobs site would appear to have the same demo data as mine.

 

I look forward to the input of any of the PS [spam-filter]’s who can contribute as to the methodology for investigating this

Link to comment
Share on other sites

  • 3 weeks later...

I noticed some speed problems on 1.4.7.x also (in some shops), the solution we have used for these shops, is to add cache feature of homefeatured, bestsellers, newproducts and other modules that load products. It's also possible to add cache feature on category pages using the same method as in blockcategoy, this makes the page slow on first load but after that the load time is below 1s.

It's important to use the update date value on the products to build the cacheid name, since this can get very long some times, we found it necessary to md5 the values. But after doing this on shops that have problems with speed, we saw great increase in performance.

 

So my tip to the team, add cache on more pages/modules :)

  • Like 2
Link to comment
Share on other sites

@prestashopmania - I compared in a timetable of 2 months. Is this not enough time ? the problem is TTFB, and as I wrote before this is not coming from the server. Ther must be a bottle-neck in the DB structure by calling many of features and many of product images for one product per time.

 

Although I compressed all mages with RIOT, all the standard png PS default are still slow by loading.

 

I use webpagetest.org, because there I have more things tested, also specially for Google optimization.

 

gtmetrix is shwoing the same problems as webpagetest is doing. TTFB, images, css and JS not well compressed. The loss from 1.4.5.1 to 1.4.8.2. is of 4-5%

Link to comment
Share on other sites

  • 1 month later...
  • 2 months later...

With so many bugs n complains about latest version, really scaring to upgrade to latest version. Currently we are using 1.3.3 so what will be better?

 

3 Options

 

1. To be on safer side by upgrading to 1.4.9

2. To upgarde to 1.5.2

3. To fresh install 1.5.2

 

I guess in OPTION 3, we will lose all existing customer, product database. Right? Plz guide. Thnx in advance!!!

 

Thanks Thomas! But that isn't quite scientific and would probably differ for everyone as well.

 

What I meant was, could you include that aspect as part of your testing suite so that over time you (and we) can look at objective speed benchmarks of different versions.

Link to comment
Share on other sites

  • 3 weeks later...

I must agree that something is terribly wrong with the speed of 1.5 ive used 1.4.3 and 1.4.7 and recently upgraded to 1.5 and spend 4 hours trying to get it working at normal speeds. Finally just had to give up and revert back.

 

Are you using Blockspecials module on your upgraded shop?? Try disabling it..

Link to comment
Share on other sites

  • 3 weeks later...

I can confirm that Prestashop 1.5.2 is DEAD SLOW !!!!

 

If you use MYISAM, it will speed up a bit, but not enough if you have more then 1000 products...

I have 8500 products: dead slow...

 

Same products in a Prestashop 1.4.9.0 shop: fast !!!!

 

Prestashop 1.5.x looks nice and has a lot of nice things but:

 

* IT"S DEAD SLOW

* NO database system change MYISAM or INNODB switch option after installation possible anymore...

 

 

So, Dear PrestaShop company, what have you done with the speed ????????!!!!

 

This will kill prestashop if this will not fixed !!!

Link to comment
Share on other sites

Is it really so dead???? This will really kill prospects of presta..............Any other shop owner who has something +ve abt PS 1.5X

 

I can confirm that Prestashop 1.5.2 is DEAD SLOW !!!!

 

If you use MYISAM, it will speed up a bit, but not enough if you have more then 1000 products...

I have 8500 products: dead slow...

 

Same products in a Prestashop 1.4.9.0 shop: fast !!!!

 

Prestashop 1.5.x looks nice and has a lot of nice things but:

 

* IT"S DEAD SLOW

* NO database system change MYISAM or INNODB switch option after installation possible anymore...

 

 

So, Dear PrestaShop company, what have you done with the speed ????????!!!!

 

This will kill prestashop if this will not fixed !!!

Link to comment
Share on other sites

Hello there,

I just have here a local test install of 1.52. with about 700 products, multi shop enabled and my process time is about 500 ms. I really can't say that this is slow. If you have trouble benchmark your config and look why it takes to long.

Best regards, tripLoad time: 461ms

Good boy! That's what I call a webserver!

  • config: 36ms
  • constructor: 0ms
  • init: 58ms
  • checkAccess: 0ms
  • setMedia: 0ms
  • postProcess: 0ms
  • initHeader: 0ms
  • initContent: 309ms
  • initFooter: 11ms
  • display: 46ms

Hook processing: 341ms / 15.55 Mb

  • displayLeftColumn: 119ms / 2.86 Mb
  • displayHome: 67ms / 1.79 Mb
  • displayHeader: 54ms / 6.62 Mb
  • displayRightColumn: 45ms / 1.57 Mb
  • displayTop: 24ms / 0.93 Mb
  • actionDispatcher: 16ms / 0.8 Mb
  • displayFooter: 11ms / 0.69 Mb
  • displayMyAccountBlock: 5ms / 0.29 Mb
  • DisplayOverrideTemplate: 0ms / 0 Mb
  • actionFrontControllerSetMedia: 0ms / 0 Mb

Memory peak usage: 25.9 Mb

  • config: 5.12 Mb
  • constructor: 0 Mb
  • init: 4.5 Mb
  • checkAccess: 0 Mb
  • setMedia: 0 Mb
  • postProcess: 0 Mb
  • initHeader: 0 Mb
  • initContent: 13.79 Mb
  • initFooter: 0.7 Mb
  • display: 0.22 Mb

Total cache size (in Cache class): 1.03 Mb

DB type: DbPDO

SQL Queries: 292 queries

Time spent querying: 119ms

Included files: 235

Size of included files: 2.68 Mb

Link to comment
Share on other sites

In /config/defines.inc.php you can enable

define('_PS_DEBUG_PROFILING_', true);

.

Please use that with caution. It might temporarely break your front or backend if there are bugs. As it says - it is for debug profiling. I would test that on a local copy and only on a live server when you are sure there are no custumers. I have just a test installation here. My overall experience is, I'll wait for the next version ;) but not because of performance issues. I thing it might even be a little faster than the 1.4.9 but it is a long time ago that i benchmarked that in comparison. No doubt with the more complexity the server specs are higher. So if you have a massive memory problem than you might run into trouble with the new version but I can not copy that this is a prestashop imminent problem - at least not with my configuration here.

Best regards, trip

Link to comment
Share on other sites

@Trip.. Thanks for this debug option. This is something new for me on PS and never noticed it.

 

I agree the first byte is 2 sec but only after removing blockspecial module in 1.5.2 installation. If I enables the same module it goes to 5 sec for first byte. In 1.4.7 I have seen the first byte coming within 1.5 sec.

 

Although this is 500ms loss but looks significant when the server is hosted outside the country in remote location due to additional network delay involved.

 

For blockspecial module I have posted in multiple threads and also raised a bug but developers are working on more critical issues right now then to work on fixing the speed related issues. The people who are directly impacted are the one`s who have upgraded their DB. Fresh install is still good.

Link to comment
Share on other sites

  • 1 month later...
  • 1 month later...

So, take a look on my debug info from store with about 1000 products.

Please let somebody explai to me.. why for god sake SQL Queries: 231 queries ?? on the homepage :-?

Memcached has been configured, APC module also installed.

Any help will be apreciated..Load time: 8.889s

You'd better run your shop on a toaster

  • config: 95ms
  • constructor: 0ms
  • init: 118ms
  • checkAccess: 0ms
  • setMedia: 0ms
  • postProcess: 0ms
  • initHeader: 0ms
  • initContent: 7.351s
  • initFooter: 9ms
  • display: 1.315s

Hook processing: 7.466s / 35.9 Mb

  • displayHome: 4.098s / 2.55 Mb
  • displayHeader: 2.241s / 27.22 Mb
  • displayTop: 486ms / 1.88 Mb
  • displayLeftColumn: 392ms / 1.32 Mb
  • displayRightColumn: 135ms / 0.22 Mb
  • actionDispatcher: 91ms / 1.03 Mb
  • headertop: 12ms / 0.13 Mb
  • displayFooter: 9ms / 1.05 Mb
  • freePosition: 3ms / 0.5 Mb
  • DisplayOverrideTemplate: 0ms / 0 Mb
  • footer_twitter: 0ms / 0 Mb
  • content_bottom: 0ms / 0 Mb
  • actionFrontControllerSetMedia: 0ms / 0 Mb
  • footer_bottom: 0ms / 0 Mb

Memory peak usage: 53.69 Mb

  • config: 10.01 Mb
  • constructor: 0 Mb
  • init: 6.69 Mb
  • checkAccess: 0 Mb
  • setMedia: 0 Mb
  • postProcess: 0 Mb
  • initHeader: 0.01 Mb
  • initContent: 33.21 Mb
  • initFooter: 1.06 Mb
  • display: 1.42 Mb

Total cache size (in Cache class): 1.03 Mb

DB type: DbPDO

SQL Queries: 231 queries

Time spent querying: 2.398s

Included files: 247

Size of included files: 2.13 Mb

Globals (> 1 Ko only): 20 Ko

  • _SERVER ≈ 9.7 Ko
  • _COOKIE ≈ 2.6 Ko

Link to comment
Share on other sites

So, take a look on my debug info from store with about 1000 products.

Please let somebody explai to me.. why for god sake SQL Queries: 231 queries ?? on the homepage :-?

Memcached has been configured, APC module also installed.

Any help will be apreciated..Load time: 8.889s

 

 

Which version are you running? VPS specifications?

Link to comment
Share on other sites

If i remember right, there where some serious speed improvements, at least the bugfix list is huge.

 

yes indeed very "serious improvement" on 1.5.3.1 :-))

 

total page load merged by firefox addon was: 58 sec !

This time I installed all shop and database on my empty machine (has clean install of ubuntu and LAMP):

 

2 x Intel® Celeron® CPU G550 @ 2.60GHz

4 GB RAM 1333MHz

 

Any ideas..?

 

Load time: 16.007s

You'd better run your shop on a toaster

  • config: 93ms
  • constructor: 0ms
  • init: 43ms
  • checkAccess: 0ms
  • setMedia: 0ms
  • postProcess: 0ms
  • initHeader: 0ms
  • initContent: 15.453s
  • initFooter: 12ms
  • display: 406ms

Hook processing: 15.494s / 20.25 Mb

  • displayHeader: 7.744s / 15.55 Mb
  • displayHome: 7.601s / 1.13 Mb
  • displayTop: 41ms / 1.14 Mb
  • displayLeftColumn: 39ms / 0.73 Mb
  • displayRightColumn: 27ms / 0.12 Mb
  • moduleRoutes: 15ms / 0.63 Mb
  • headertop: 13ms / 0.08 Mb
  • displayFooter: 12ms / 0.63 Mb
  • freePosition: 3ms / 0.23 Mb
  • footer_twitter: 0ms / 0 Mb
  • actionDispatcher: 0ms / 0 Mb
  • DisplayOverrideTemplate: 0ms / 0 Mb
  • content_bottom: 0ms / 0 Mb
  • actionFrontControllerSetMedia: 0ms / 0 Mb
  • footer_bottom: 0ms / 0 Mb

Memory peak usage: 30.82 Mb

  • config: 6.37 Mb
  • constructor: 0 Mb
  • init: 3.47 Mb
  • checkAccess: 0 Mb
  • setMedia: 0 Mb
  • postProcess: 0 Mb
  • initHeader: 0.01 Mb
  • initContent: 18.69 Mb
  • initFooter: 0.64 Mb
  • display: 0.89 Mb

Total cache size (in Cache class): 0.57 Mb

 

DB type: DbPDO

SQL Queries: 247 queries

Time spent querying: 14.618s

Included files: 250

Size of included files: 2.18 Mb

Globals (> 1 Ko only): 12 Ko

  • _SERVER ≈ 5.8 Ko
  • _COOKIE ≈ 2.4 Ko

Edited by intro (see edit history)
Link to comment
Share on other sites

As the benchmark states

  • displayHeader: 7.744s / 15.55 Mb
  • displayHome: 7.601s / 1.13 Mb

You should disable modules/hooks in the header and homepage untill you find out which of them takes so long.

Usually wenn you have define('_PS_DEBUG_PROFILING_', true); you get very detailed statistics on the database queries and how long every query took.

 

The often cited define('_PS_MODE_DEV_', true); can give hints if modules are not working at all. I can assure you that my 1.5.3.1. runs (under optimal conditions) only minimal slower than the 1.4.9.

 

 

Your processing is unnaturally high but you should keep in mind afaik the _PS_DEBUG_PROFILING_ bypasses the caching of mysql. I for example have a module that takes maybe 2 seconds for a not so optimal table sorting. But as mysql caches the query my actual process time stays under one second. That just said that the synthetic benchmark might not reflect the true performance with sql-chaching on. But anyway.. your case seems serious so try to find out if a special module is the stopper or the server config / ram etc. pp. is the cause.

Best regards, trip

Link to comment
Share on other sites

Ok, I try to help debugging the issue.

Here are my stats on a local installation 1.53 (upgraded shop - about 2000 product in 4 languages). with 50 products on a category view:

Load time: 497ms
Good boy! That's what I call a webserver!
config: 48ms
constructor: 0ms
init: 38ms
checkAccess: 0ms
setMedia: 0ms
postProcess: 0ms
initHeader: 0ms
initContent: 299ms
initFooter: 19ms
display: 93ms

So I guess with the right config it is performing quite well. Under some circumstances we notices it does not work very good so you have to find out the problem. I can not copy a general diagnosis like it is not performing well.

Here is a bench of my test 1.5.3. shop (best of 3 runs) http://www.webpagetest.org/result/130220_A8_5QH/1/details/ ... it loads in about 2 seconds... server located in germany ... benchmark from Dulles

Version 1.4.9. best of 3 runs -> 2.7. seconds... http://www.webpagete.../130220_M8_5TA/

 

Regards, trip

Edited by Trip (see edit history)
  • Like 1
Link to comment
Share on other sites

Trip you are just great.......thnx for sharing all these

 

Ok, I try to help debugging the issue.

Here are my stats on a local installation 1.53 (upgraded shop - about 2000 product in 4 languages). with 50 products on a category view:

Load time: 497ms
Good boy! That's what I call a webserver!
config: 48ms
constructor: 0ms
init: 38ms
checkAccess: 0ms
setMedia: 0ms
postProcess: 0ms
initHeader: 0ms
initContent: 299ms
initFooter: 19ms
display: 93ms

So I guess with the right config it is performing quite well. Under some circumstances we notices it does not work very good so you have to find out the problem. I can not copy a general diagnosis like it is not performing well.

Here is a bench of my test 1.5.3. shop (best of 3 runs) http://www.webpagete..._5QH/1/details/ ... it loads in about 2 seconds... server located in germany ... benchmark from Dulles

Version 1.4.9. best of 3 runs -> 2.7. seconds... http://www.webpagete.../130220_M8_5TA/

 

Regards, trip

Link to comment
Share on other sites

Ok here goes one more for the statistic fans :) ... Testsetup Apache Bench --10 concurrent connection 100 requests:

as found here http://www.petefreitag.com/item/689.cfm

ab -n 100 -c 10 http://www.example.com/

1.4.9.

Concurrency Level:	  10
Time taken for tests:   8.078 seconds
Complete requests:	  100
Failed requests:	    0
Write errors:		   0
Non-2xx responses:	  100
Total transferred:	  78016 bytes
HTML transferred:	   0 bytes
Requests per second:    12.38 [#/sec] (mean)
Time per request:	   807.775 [ms] (mean)
Time per request:	   80.777 [ms] (mean, across all concurrent requests)
Transfer rate:		  9.43 [Kbytes/sec] received
Connection Times (ms)
		  min  mean[+/-sd] median   max
Connect:	   28   87 155.4	 32    1030
Processing:   533  616 222.7    564    1643
Waiting:	  533  616 222.7    564    1643
Total:	    562  702 288.1    602    1850
Percentage of the requests served within a certain time (ms)
 50%    602
 66%    635
 75%    648
 80%    664
 90%    827
 95%   1594
 98%   1836
 99%   1850
100%   1850 (longest request)

 

1.5.3.1

Concurrency Level:	  10
Time taken for tests:   7.025 seconds
Complete requests:	  100
Failed requests:	    0
Write errors:		   0
Non-2xx responses:	  100
Total transferred:	  30400 bytes
HTML transferred:	   0 bytes
Requests per second:    14.24 [#/sec] (mean)
Time per request:	   702.471 [ms] (mean)
Time per request:	   70.247 [ms] (mean, across all concurrent requests)
Transfer rate:		  4.23 [Kbytes/sec] received
Connection Times (ms)
		  min  mean[+/-sd] median   max
Connect:	   26   30   4.0	 29	  47
Processing:   532  656 320.4    561    1696
Waiting:	  532  656 320.4    561    1696
Total:	    561  687 323.3    590    1740
Percentage of the requests served within a certain time (ms)
 50%    590
 66%    597
 75%    600
 80%    601
 90%    612
 95%   1723
 98%   1740
 99%   1740
100%   1740 (longest request)

 

These tests were made from a distance of about 9000 kilometers so that 1.5.3.1 is performing better again might be coincidence. What I want to point out is, that the performance is comparable good assuming you have a server that has a little juice ;).

 

Best regards, trip

  • Like 1
Link to comment
Share on other sites

Server listed below is a dedicated PS server - we have 11 sites running PS 1.5.2.0 and the loads are very high, I've seen them as high as 90.0

 

I Like PS but I am not impressed on the resources it needs to run, however I do have one site running 1.5.3.1 and it seems as slow as the others do.

 

 

Metric Status InterWorx-CP Version: InterWorx-CP v4.11.3 [unlimited Domain] SiteWorx Accounts: 11 / Unlimited Used Distribution: CentOS release 6.3 (Final) Operating System: Linux 2.6.32-17-pve (SMP) CPU: [4x] 2200.08Mhz Dual-Core AMD Opteron Processor 2214 HE (1024 KB Cache) System Memory: 7168.00 MB System Uptime: 35 days 19 hours 28 minutes Load Average: 33.92 27.57 17.68

Link to comment
Share on other sites

Hi anves. Unfortunately I can not access your stats. So what exactly does produce the load MySQL? Apache? I am not an expert but I did some basic optimisations on the database to dedicate more RAM for caching to MySql. Unused RAM is wasted ram. The standard MySQL config is the minimal config for really slow/old systems.

You can use scripts like

  1. mysqltuner.pl
  2. tuning-primer.sh

to adjust the settings. I prefer mysql tuner. It shows you also optimisation tips. As I remember I adjusted

innodb_buffer_pool_size for innodb tables.

 

MyISAM:

key_buffer

query_cache_limit

query_cache_size

table_cache

 

I have about 20 running sites on my server. 3 Shop Systems, Big WIKI sites and cpu load is by an average of about 2%. When you have such high loads find out which processes consume so much cpu. Just a shot in the dark. I assume that the server is swapping due a misconfigured apache. I have pretty much the default config

<IfModule mpm_prefork_module>
StartServers		  5
MinSpareServers	   5
MaxSpareServers	  10
MaxClients		  150
MaxRequestsPerChild  1000
</IfModule>

but I know that for example having to much or to low spare servers has a huge impact on the overall performance. When apache can not handle enough requests and/or starts spawning new servers than you get high cpu loads and a miserable performance. Mis-/and or Bad configuration here can easily crash your server.

The keep alive setting can also be a huge factor. If they are misconfigured and depending on the traffic the server might wait to long for a new client connection and therefore waste valuable resources.

ATM I use:

KeepAlive On
#
# MaxKeepAliveRequests: The maximum number of requests to allow
# during a persistent connection. Set to 0 to allow an unlimited amount.
# We recommend you leave this number high, for maximum performance.
#
MaxKeepAliveRequests 1000
#
# KeepAliveTimeout: Number of seconds to wait for the next request from the
# same client on the same connection.
#
KeepAliveTimeout 3

Worst case your server runs out of ram because it has spawned to many servers to handle the request. If it starts swapping to harddrive a total crash might not be far away and your server runs at limit. I had that case once on my own and I had to shutdown the site causing the problems.

 

Afaik there are no general recommendations.You have to change these settings carefully in small steps to find out the right balance. As I said, I only have basic knowledge if there are some pros here feel free to correct me. Maybe there are other/better tools to find out bottle necks or checkout the config? From what I experienced the most annoying part is the apache tuning :)

 

By the way. I would like to recommend Munin for server monitoring. You might have to google a bit how to enable the apache and mysql monitor plugins but it is definetly worth to get a good overview on you server performance.

 

Best regards, trip

Edited by Trip (see edit history)
Link to comment
Share on other sites

Nice to see some good 1.5.3.1. results..

 

Server settings/software versions can make or break it..

an update to one of te latest php and mysql brought me 15% increase in performance on a vps.. without any adjustments.

just a "basic" image with a few tweaks in the settings and eaccelerator installed.

 

now working on ningx as full apache replacement to see what that does.

Link to comment
Share on other sites

  • 1 month later...

Hi

I have just started testing a version of 1.5 as a final release seems to be getting closer and configured a very basic shop with the demo products and was amazed at how slow it appears compared with my current 1.4.7 installation. This was all done on a local test machine but should give a good comparison.

 

Without going into the many additional ways of improving speed beyond those easily achieved in the BO UI, has anyone else noticed the speed decrease in 1.5 and does anyone know if anything will be done before final release.

 

Thanks

 

 

I HAVE SOLVE IT!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

o do that, navigate to the "Preferences -> Images" page and click "Move Images" button under the "Move Images" section. It will take a while. After it's completed, select "No" for "Use the legacy image filesystem" option under the "Product Images" section and save your configuration.

That should fix your issues.

Best Regards.

Link to comment
Share on other sites

Wow, Changing the CCC in Preformance Made a Huge Diffrence

with the initial and total load time

 

i tested many variations

endlessely over and over again

 

this is what made my site preform the best

arrow_in.png CCC (Combine, Compress and Cache)

CCC allows you to reduce the loading time of your page. With these settings you will gain performance without even touching the code of your theme. Make sure, however, that your theme is compatible with PrestaShop 1.4+. Otherwise, CCC will cause problems.

Smart cache for CSS

✔Use CCC for CSS.

Keep CSS as original

 

Smart cache for JavaScript

✔Use CCC for JavaScript.

Keep JavaScript as original

 

Minify HTML

Minify HTML after "smarty compile" execution.

✔Keep HTML as original

 

Compress inline JavaScript in HTML

Compress inline JavaScript in HTML after "smarty compile" execution

✔Keep inline JavaScript in HTML as original

 

Apache optimization

on ✔

enabled.gif

This will add directives to your .htaccess file which should improve caching and compression.

 

it seems the minigfy and compress made the intial bytes take longer waiting on server because prestashop was processing the file before spiting it..

 

minify html - added .5 ms to initial load time

compress added 1 ms to initial load time

 

hope this helps, my site is at 3.25 sec now

and initial load file load is 1.5 sec, anyone have otherways to improve it

Edited by hackia (see edit history)
  • Like 1
Link to comment
Share on other sites

It is really deadly slow compare to 1.4.6 version. I install the 1.5.2 at a dedicated server using memcache 640M; And it's even 2 seconds slower than the magento on the same server!! And 1.4.6 is normal faster than magento before upgrade to 1.5.2!!! Is there an solution? Otherwise, I have to consider using opencart, which is very light-weighted and faster for small shop with limited sku

Link to comment
Share on other sites

  • 1 month later...

Since I updated to version 1.5 has been a hassle to find the reason for this slowness in backoffice.
The appearance is very pretty but it works much better with the layout of version 1.4
Prestashop must lost the fuc.. habit of transparencies on backoffice, occupies many resources.

Dedicated server quad core with 4G Ram
I was wrong: it`s dual core and 2G Ram

Before modification:
My test,
Customers tab>Set Display "300"
It took
PrestaShop™ 1.5.3.1
Load time:14.320s

My modifications:
Tab>advanced parameters>Performance

Use CCC for CSS. Yes
Use CCC for JavaScript. Yes
Keep HTML as original Yes
Keep inline JavaScript in HTML as original Yes
Apache optimization Yes
(Caching) Use cache NO

And now the miracle,
In modules> uninstall the following modules
Analytics for ecommerce
Best categories
Best manufacturers
Best suppliers
Best vouchers
Catalog evaluation
Geolocation
Páginas não encontradas
Shop search
Software
Visitors online
Visitors origin
And other crap you do not really need

VERY IMPORTANT
In module statsdata
Save page views for each customer NO
Save global page views NO
Plug-ins detection NO


In tab Stats>Referrers>Save direct traffic> Set to NO

AFTER MODIFICATION

Astonishing!!! :D

PrestaShop™ 1.5.3.1
Load time: 2.881s

After numerous attempts and this experience was the best configuration.
Problem solved

_______________________________________________________________________

I performed the upgrade to 1.5.4.1 yesterday and decided to do the test again.
Well... it was a bit disappointing! :(


PrestaShop™ 1.5.4.1
Load time: 7.951s
_______________________________________________________________________

Correction:

After the upgrade I had a custom module did not work in proper conditions.
After the uninstall this module,

PrestaShop™ 1.5.4.1
Load time: 2.936s


It`s ok :)

_______________________________________________________________________

 

New update.

I upgraded to version PrestaShop™ 1.5.6.0

And most importantly, I changed my server, advise everyone to try other internet servers.

 

My test,
Customers tab>Set Display "300"
It took
PrestaShop™ 1.5.6.0
Load time::0.835s

 

Orders tab>Set Display "300"

 

It took
PrestaShop™ 1.5.6.0
Load time::0.735s

 

Catalog > Products > Edit(product)

PrestaShop™ 1.5.6.0
Load time:0.169s

 

Save product

PrestaShop™ 1.5.6.0
Load time:0.169s
 
As you can see everything below 1 second, as said above "advise everyone to try other internet servers"
Everything working spectacularly fast.
 
Until next time  :D
Edited by 1117 (see edit history)
  • Like 1
Link to comment
Share on other sites

Nice! I also found (as have others) that the included Fedex module slows everything down. Especially the Add to Cart. It's unfortunate because I really need Fedex, but I'm waiting for the next version. But with all the extraneous modules uninstalled, I'm finding that 1.5.4 is pretty much as fast as my large shop on 1.4.10.

Link to comment
Share on other sites

  • 10 months later...
×
×
  • Create New...