python - Django - Cache vs QuerySet Filter for Server Efficiency? -

which 1 of these 2 more efficient on server resources?

lets have 1 "main" queryset use , have 3 additional filters display filtered results.

option 1:

q = entry.objects.filter(filter='filter') = q.filter(filter=a) b = q.filter(filter=b) c = q.filter(filter=c) 

option 2.

lets assume can cache filtered results memcached can either run option 1 or can following:

q = entry.objects.filter(filter='filter') = cache.get('key_a') b = cache.get('key_b') c = cache.get('key_c')     

is correct assumption option 1 1 database query , the filters applied django? or wrong , running option 1 database hit 4 times?

now if option 1 = 1 database hit, more efficient process a, b, c, filters django (or rather web server) versus grabbing results cache?

i cache q results technically have these 2 options:

option 1a:

get q cache , apply queryset filters

q = cache.get('key_q') = q.filter(filter=a) b = q.filter(filter=b) c = q.filter(filter=c) 

option 2a:

get cache

q = cache.get('key_q') = cache.get('key_a') b = cache.get('key_b') c = cache.get('key_c')  

what correct, better approach django 1.8 running standard linux/apache/mysql stack on aws?

is correct assumption option 1 1 database query , the filters applied django?

no. filter(lookup=value) calls used build sql where clause.

or wrong , running option 1 database hit 4 times?

it's bit more complex... none of 4 statements in option 1 result in database hit, until start iterating on querysets. yes, have (at least - cf below) 1 query per queryset, of course.

so mostly, caching "unevaluated" queryset pretty useless since @ point nothing has been loaded database.

nb: "at least one" because can have n+1 queries if start following foreign keys on model instances (unless built queryset select_related clause).