Before making any change on your PrestaShop instance, it is highly recommended to perform some benchmarks.
The idea behind this is to know what your baseline performance is, so that you can make sure that your changes are increasing your shop’s performance and not the other way around.
If you’ve done no benchmark beforehand, there is no way to ensure you did not decrease your performances in the end.
Keep in mind that the surest way to tune your shop is:
Performance tuning can be a tricky thing, there are plenty of factors that can slow down an application.
It can come from your disks, lack of memory, network capacity, CPU: wherever you can think of.
Then, the challenge is to find your current bottleneck (could be disks access), remove it and then find the next one (could be memory).
There’s no way to enumerate all the performance issues we’ve encountered, but let’s review the more common ones and tackle them.
PHP tuning is very important for your application’s performance, whether you’re running PrestaShop or any other PHP software.
First, try to use PHP >=7 when possible. Hard work has been done on performance starting on this version and it will provide a nice speed boost to your shop!
If you’re using PHP-FPM (which should be the case for most “modern” installations), you have to check the pool configuration.
It’s usually stored in the file /etc/php/7.x/fpm/pool.d/www.conf.
Inside this file, the most important setting is the pm.max_children setting. It must be greater than the max number
of concurrent users you want to simulate during the bench.
Use the following settings to optimize the performances:
[Date]
date.timezone = UTC
[Session]
session.auto_start = Off
[PHP]
short_open_tag = Off
display_errors = Off
magic_quotes_gpc = off
; Increase this value if you are able to do it
memory_limit = 512M
max_execution_time = 300
max_input_time = 300
upload_max_filesize = 20M
post_max_size = 22M
; Increase this value if you work with products with a lot of combinations
max_input_vars = 20000
allow_url_fopen = on
It is well known that PHP does not manage file systems very well.
That’s why there are plenty of useful tuning options to avoid accessing the file system constantly.
At each file access, by default, PHP will first check that the file is still there, causing plenty of lstat system calls. Potentially thousands.
PHP provides an option to store this information in cache, so that it can avoid repeating those system calls constantly:
[PHP]
realpath_cache_size = 4096K
realpath_cache_ttl = 600
Keep in mind that this configuration is NOT compatible with other parameters, such as open_basedir and safe_mode directives.
Finally, if you’re using a NAS or any other network storage solution to store your files (in case of horizontal scaling, for example), it is highly recommended to enable this setting.
Because file system tuning is a never ending story, not only you can cache the files’ path, but also its contents.
Good news is, OPCache will not only store your PHP files in memory, but it will store their bytecode, meaning the already compiled application, in a shared memory, available to all the application’s calls:
[opcache]
opcache.enable=1
opcache.enable_cli=0
opcache.memory_consumption=256
opcache.interned_strings_buffer=32
opcache.max_accelerated_files=16229
opcache.max_wasted_percentage=10
opcache.revalidate_freq=10
opcache.fast_shutdown=1
opcache.enable_file_override=0
opcache.max_file_size=0
If you can manage it, here are few options you could configure:
opcache.validate_timestamps=0
opcache.revalidate_path=0
Keep in mind if you deactivate validate_timestamps OPCache will never update your code unless you let it know explicitly (either through internal functions or by restarting the web server).
Also, your favorite ecommerce project made sure it’s fully compatible with OPCache.
Isn’t it nice?
The class loader used while developing the application is optimized to find new and changed classes. On production servers, PHP files should never change, unless a new application version is deployed. That’s why you can optimize Composer’s autoloader to scan the entire application once and build a “class map”, which is a big array of all the classes locations, stored in vendor/composer/autoload_classmap.php.
Execute this command to generate the class map (and make it part of your deployment process):
composer dump-autoload --optimize --no-dev --classmap-authoritative
--optimize: dumps every PSR-0 and PSR-4 compatible class used in your application;--no-dev: excludes the classes that are only needed in the development environment (e.g. tests);--classmap-authoritative: prevents Composer from scanning the file system for classes that are not found in the class map.If you’re using PHP-FPM, you should be able to use Apache mpm_event. Using the following configuration (to set in the mpm_event.conf file) should allow you to test up to 400 concurrent users:
ServerLimit 16
MaxClients 400
StartServers 3
ThreadLimit 64
ThreadsPerChild 25
MaxRequestWorkers 400
MaxConnectionsPerChild 0
Though we’re discussing it just now, MySQL tuning is just as important.
Our intention here is to optimize the instance’s throughput by adding caches.
As for PHP, it allows the service to work as much as possible in memory and avoid disk access, hence reducing latency.
These parameters allow better cache information for further reuse, first by enabling it, then by increasing its size. Again, the idea is to keep the query results in memory rather than looking them up to the (higher latency) hard drive.
As always, these values should be adjusted to your own environment, you probably won’t need a host_cache_size of 1000.
query_cache_limit = 128K
query_cache_size = 32M
query_cache_type = ON
table_open_cache = 4000
thread_cache_size = 80
host_cache_size=1000
Buffering is almost another word for caching.
So we work here with the memory area that holds cached data for InnoDB tables, indexes, and other auxiliary buffers, etc..
Again, these options should be adapted to your shop.
Setting up the innodb_buffer_pool_size to 1G may be too much for your SQL instance, just make sure you have enough memory according to the value you configure.
The important thing is, if possible, to set innodb_buffer_pool_size to a greater value than your database size.
read_buffer_size = 2M
read_rnd_buffer_size = 1M
join_buffer_size = 2M
sort_buffer_size = 2M
innodb_buffer_pool_size = 1G
Some other parameters to increase the MySQL performance, such as disabling the performance schema (used for monitoring), memory tables and enhancing GROUP BY queries.
performance_schema = OFF
max_heap_table_size = 32M
tmp_table_size = 32M
Before launching the benchmark, and after importing the data, it’s always great to launch an ANALYZE TABLE on all your database by running on your server:
mysqlcheck -a -A -uroot -pyour_password
A CDN (Content Delivery Network) is a kind of proxy that will cache your static files and serve them instead of your own server.
Hence, a CDN will reduce drastically the amount of hits made to your server.
There are plenty of CDN providers, Cloudflare is a good and free one. You can use as well to minify your page or to easily enable SSL on your site.
Just one slight consideration about CDNs though: any modification done to your shop’s static assets, such as images, css and the like, may not be immediately available once behind a CDN. When those files are cached, if you modify them on your server, you will need to wait for the cache expiration (often configurable) before seeing the modification live - or invalidate part or all your cache. Most providers offer such features.
In the performances tab of your PrestaShop (Advanced Parameters > Performance):
In the file config/defines.inc.php:
_PS_MODE_DEV_ to false.if (!defined('_PS_MODE_DEV_')) {
define('_PS_MODE_DEV_', false);
}
_PS_DEBUG_PROFILING_ to false.if (!defined('_PS_DEBUG_PROFILING_')) {
define('_PS_DEBUG_PROFILING_', false);
}
See the Taking Care of PrestaShop page section.
PrestaShop can be optimized, but the other systems running on it should be optimized too! Some modules can slow your shop down especially if they are hooked on critical Front Office pages. You should benchmark and profile the modules installed on your shop as they are not always fine-tuned for optimization.
Things you might look for in these modules
*For example some modules do rely on Smarty cache, but Smarty cache badly configured can result in hundred of template files being scattered on the server, producing way too much I/O.
A simple setup for MySQL replication strategy is a single instance for WRITE requests and multiple instances for READ requests. Because the Front Office actually does not perform a lot of write requests this can boost your Front office performances.