python - Django: Extremely Large Queryset Slow to Execute (non-linear retrieval time) -

i having django (oracle backend, cx_oracle 5.2.1) execute query simple

   select x.y, x.z x x.a = 'foo' , x.b = 'bar' 

both x.a , x.b indexed (db_index=true)

the page use calculates summary statistics on various records using pandas, data sizes quite large.

i have 1 page loading subset of ~15000 records. takes ~2 seconds load

the exact same page, when using different timeframe, using same type of query loads ~50000 records. takes ~30 seconds load. i've benchmarked performance, , pandas not bottleneck.

the simple

my_queryset.all().values(('y', 'z'))  

takes overwhelming majority of time (~90%)

is there reason why is, or way speed up?

of note, when executing these queries in database console, both fast. seems django orm has issues.