Hi~
I want to use aod_sketch in our company project. But i meet performance issues that use aod_sketch.
I run this query in the PostgreSQL with version postgresql10-10.9-1PGDG.rhel7.x86_64.rpm. And the machine is AWS ec2, Redhat, C5.9Xlarge ( 36 vCPUs, 72G memory).
And the data size is very small.
Data size: Just one 4k sketch data intersection one 4K sketch data. But the SQL runs for almost 10 seconds.
Are there any performance know issues ? Or other parameters that can improve the performance. Did you meet the same problem?
SQL like this:
with
a as(select sk from test_ds_sketch where id ='3113'), # just one record
b as ( select sk from test_ds_sketch where id='2662') # just one record
,
c as (select 171213 as network_id,aod_sketch_get_estimate(aod_sketch_intersection(a.sk,b.sk))/aod_sketch_get_estimate(a.sk)
as ratio from a,b),
select * from c
Hi~ I want to use aod_sketch in our company project. But i meet performance issues that use aod_sketch. I run this query in the PostgreSQL with version postgresql10-10.9-1PGDG.rhel7.x86_64.rpm. And the machine is AWS ec2, Redhat, C5.9Xlarge ( 36 vCPUs, 72G memory). And the data size is very small. Data size: Just one 4k sketch data intersection one 4K sketch data. But the SQL runs for almost 10 seconds. Are there any performance know issues ? Or other parameters that can improve the performance. Did you meet the same problem? SQL like this: