How do I optimize this query with whereHas? - php

Question

I was using this query to filter out stores with city and categories. It was working fine when I had around 1000 records in stores table. Now, when I have 5000 records it takes around 3-10 seconds to generate result.
A store belongs to multiple categories in my case.
How can I optimize this query using Eloquent orm or DB::raw()?
$stores = Store::where('city_id', $city_id)
->where('disabled', '0')
->whereHas('categories', function($q) use ($category_id){
$q->where('category_id', '=', $category_id);
})
->orderBy('rating','DESC')->paginate(10);

Answer 1

I solved my problem using whereRaw as DB::raw() or DB::select() can not paginate() the collection.
Problem:
Execution time: 11.449304103851s
city_id = 6 & $category_id = 1
$stores = Store::where('city_id', $city_id)
->where('disabled', '0')
->whereHas('categories', function($q) use ($category_id){
$q->where('category_id', '=', $category_id);
})
->orderBy('rating','DESC')->paginate(10);
Solution:
Execution time: 0.033660888671875s
city_id = 6 & $category_id = 1
$stores = Store::
where('city_id', $city_id)
->where('disabled', '0')
->whereRaw('stores.id in (select store_id from store_categories_pivot where category_id = ?)', [$category_id])
->orderBy('rating','DESC')
->paginate(10);

Answer 2

You could try running:
$stores = Store::where('city_id', $city_id)
->where('disabled', '0')
->leftJoin('categories','categories.store_id','=', 'stores.id')
->where('category_id', $category_id)
->orderBy('rating','DESC')->paginate(10);
and now verify your time execution. But you might need to add extra changes to such query because we don't know exact tables structure and how data is organized in them.
If it doesn't help you should get the query that is executed (exact query) and then run
EXPLAIN your_query
in Database Tool to show you what exactly is happening and whether do you really have indexes on everything that is needed.
Looking at your query you should probably have indexes for stores for columns:
city_id
disabled
rating
and for categories you should have indexes for columns:
category_id
store_id
or for some combinations of those columns.