Performance with 100k products

Hi,

I would like to ask about performance related to search in community edition. I made some test on my local environment (docker / ubuntu / 8GB / SSD / i7):
– Added 100 000 products to catalog
– Use frontend search with some keyword (returned 98 000 products, 25 per page)

I noticed that retrieving results took 36 secounds. Only 2 queries had been executed long time (select found items, and count found items). The same action with standard demo catalog takes abount 2 seconds. My mysql has a little bit changed configuration (innodb_buffer_pool_size = 6GB etc.). I know that my local env is not the same like some production env, but I am afraid that performance lack will be very similar.

Could you tell me:
1) Have you done some performance test for product catalog with 100k products?
2) Using application with 100k products will be satisfied for customer (performance, what will be average respons time)?
2) Should I think about enterprise version for eshop with that big catalog?

I am asking because I wouldn’t want to improve performance if we made application for our client.

Thanks for answer!

Topic

[MichaelKeymaster]

Yes, we’ve done some performance testing on various catalog sizes (starting with 10k up to 1M products) and are now testing even larges catalog size (close to 10M products).
We recommend to setup the deployment configuration in a way that the page response time never exceeds 2 second, and the system should respond even faster (under 200ms) for most pages (homepage, category pages, product view pages, landing pages, etc.).

It might be challenging to achieve that on the community edition, because all user activity will be generating load on the same database. We support Elasticsearch in the enterprise edition, which is more scalable and allows both for better performance and quality of the search results.

[tjurczakParticipant]

Thanks Michael,

I understand that you have done tests only in the enterprise edition, but is it possible to achieve response time bellow 3 seconds in community edition for 100k products in search on frontend? Have you tried this? What machine had you used?

[JanParticipant]

Simple search in 100 000 products by one user should works better than 36s, and don’t need Elastic search and ostgreSQL – looks like not opimized.

[Author]

Replies