The below SQL query runs for almost 3 mins. All the 3 tables (C_table, E_table, and V_table) are external tables and are partitioned on the basis of C_id, dt. Each CTE returns about 9000 columns and there are about 40 million records in each table. Is there any way I can optimize this query?
I tried changing the order by sequence, but it is still the same.
WITH S AS
(SELECT S.C_id AS C_id,
S.dt AS date,
S.CR_id AS CR_id,
sum(AI) AS AI,
sum(UE) AS UE,
sum(UW) AS UW,
cast(sum(VI) AS double) / cast(sum(AI) AS double) AS VR
FROM "db_name"."C_table" S
WHERE S.C_id IN (10422,10421,10420,10418,10417,10416,10415,10414)
AND S.dt
BETWEEN date '2021-01-1'
AND date '2021-03-29'
GROUP BY S.C_id,S.dt, S.CR_id ) ,
E AS
(SELECT E.C_id AS C_id,
E.dt AS date,
E.CR_id AS CR_id,
sum(UIEO) AS UIEO
FROM "db_name"."E_table" E
WHERE E.C_id IN (10422,10421,10420,10418,10417,10416,10415,10414)
AND E.dt
BETWEEN date '2021-01-1'
AND date '2021-03-29'
GROUP BY E.C_id,E.dt, E.CR_id ) ,
V AS
(SELECT V.C_id AS C_id,
V.dt AS date,
V.CR_id AS CR_id,
cast(sum(V_cmplt) AS double) / cast(sum(V_strt) AS double) AS VCR
FROM "db_name"."V_table" V
WHERE V.C_id IN (10422,10421,10420,10418,10417,10416,10415,10414)
AND V.dt
BETWEEN date '2021-01-1'
AND date '2021-03-29'
GROUP BY V.C_id,V.dt, V.CR_id )
SELECT cp.name AS C_name,
ct.name AS CR_name,
S.date,
S.AI,
((cast(S.UE AS double) + cast(E.UIEO AS double)) / cast(S.UW AS double)) AS ER,
V.VCR,
S.VR
FROM S
JOIN V
ON V.C_id = S.C_id
AND V.date = S.date
AND V.CR_id = S.CR_id
JOIN E
ON E.C_id = S.C_id
AND E.date = S.date
AND E.CR_id = S.CR_id
JOIN "db_name"."C_lookup" cp
ON cp.id = S.C_id
JOIN "db_name"."CR_lookup" ct
ON ct.id = S.CR_id
Any help is appreciated.