jprdonnelly/impala
1
0
Fork 0
mirror of https://github.com/apache/impala.git synced 2026-02-03 00:00:40 -05:00
Code Issues Projects Releases Wiki Activity
Files
13bbff4e4e5fc5d459cc6f7a5512f84ceba897cd
impala/testdata/workloads/functional-planner/queries/PlannerTest/multiple-distinct-materialization.test
Tim Armstrong e5777f0eb8 IMPALA-8834: Short-circuit partition key scan 
This adds a new version of the pre-existing partition
key scan optimization that always returns correct
results, even when files have zero rows. This new
version is always enabled by default. The old
existing optimization, which does a metadata-only
query, is still enabled behind the
OPTIMIZE_PARTITION_KEY_SCANS query option.

The new version of the optimization must scan the files
to see if they are non-empty. Instead of using metadata
only, the planner instructs the backend to short-circuit HDFS
scans after a single row has been returned from each
file. This gives results equivalent to returning all
the rows from each file, because all rows in the file
belong to the same partition and therefore have identical
values for any columns that are partition key values.

Planner cardinality estimates are adjusted accordingly
to enable potentially better plans and other optimisations
like disabling codegen.

We make some effort to avoid generated extra scan ranges
for remote scans by only generating one range per remote
file.

The backend optimisation is implemented by constructing a
row batch with capacity for a single row only and then
terminating each scan range once a single row has been
produced.  Both Parquet and ORC have optimized code paths
for zero slot table scans that mean this will only result
in a footer read. (Other file formats still need to read
some portion of the file, but can terminate early once
one row has been produced.)

This should be quite efficient in practice with file handle
caching and data caching enabled, because it then only
requires reading the footer from the cache for each file.

The partition key scan optimization is also slightly
generalised to apply to scans of unpartitioned tables
where no slots are materialized.

A limitation of the optimization where it did not apply
to multiple grouping classes was also fixed.

Limitations:
* This still scans every file in the partition. I.e. there is
  no short-circuiting if a row has already been found in the
  partition by the current scan node.
* Resource reservations and estimates for the scan node do
  not all take into account this optimisation, so are
  conservative - they assume the whole file is scanned.

Testing:
* Added end-to-end tests that execute the query on all
  HDFS file formats and verify that the correct number of rows
  flow through the plan.
* Added planner test based on the existing test partition key
  scan test.
* Added test to make sure single node optimisation kicks in
  when expected.
* Add test for cardinality estimates with and without stats
* Added test for unpartitioned tables.
* Added planner test that checks that optimisation is enabled
  for multiple aggregation classes.
* Added a targeted perf test.

Change-Id: I26c87525a4f75ffeb654267b89948653b2e1ff8c
Reviewed-on: http://gerrit.cloudera.org:8080/13993
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
2020-05-20 23:03:23 +00:00

1115 lines
35 KiB
Plaintext
Raw Blame History

# A subset of aggregation classes are materialized. No group by.
select a, c, e1, e2, e3 from
(select count(distinct tinyint_col) a, avg(distinct smallint_col) b,
count(distinct int_col) c, avg(distinct bigint_col) d,
min(float_col) e1, max(float_col) e2, sum(double_col) e3
from functional.alltypes) v
---- PLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,5,7,10,11) = 2, count(tinyint_col)), aggif(valid_tid(2,5,7,10,11) = 7, count(int_col)), aggif(valid_tid(2,5,7,10,11) = 11, min(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, max(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, sum(double_col))
| row-size=32B cardinality=1
|
02:AGGREGATE [FINALIZE]
| Class 0
| output: count(tinyint_col)
| Class 1
| output: count(int_col)
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| row-size=32B cardinality=3
|
01:AGGREGATE
| Class 0
| group by: tinyint_col
| Class 1
| group by: int_col
| Class 2
| output: min(float_col), max(float_col), sum(double_col)
| row-size=21B cardinality=21
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=17B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,5,7,10,11) = 2, count(tinyint_col)), aggif(valid_tid(2,5,7,10,11) = 7, count(int_col)), aggif(valid_tid(2,5,7,10,11) = 11, min(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, max(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, sum(double_col))
| row-size=32B cardinality=1
|
07:AGGREGATE [FINALIZE]
| Class 0
| output: count:merge(tinyint_col)
| Class 1
| output: count:merge(int_col)
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| row-size=32B cardinality=3
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| Class 0
| output: count(tinyint_col)
| Class 1
| output: count(int_col)
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| row-size=32B cardinality=3
|
05:AGGREGATE
| Class 0
| group by: tinyint_col
| Class 1
| group by: int_col
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| row-size=21B cardinality=21
|
04:EXCHANGE [HASH(CASE valid_tid(1,6,11) WHEN 1 THEN murmur_hash(tinyint_col) WHEN 6 THEN murmur_hash(int_col) WHEN 11 THEN 0 END)]
|
01:AGGREGATE [STREAMING]
| Class 0
| group by: tinyint_col
| Class 1
| group by: int_col
| Class 2
| output: min(float_col), max(float_col), sum(double_col)
| row-size=21B cardinality=21
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=17B cardinality=7.30K
====
# A subset of aggregation classes are materialized. A subset of aggregation functions
# of the surviving classes are materialized. No group by.
select a1, c2, e2 from
(select count(distinct tinyint_col) a1, avg(distinct tinyint_col) a2,
count(distinct smallint_col) b1, avg(distinct smallint_col) b2,
count(distinct int_col) c1, avg(distinct int_col) c2,
count(distinct bigint_col) d1, avg(distinct bigint_col) d2,
min(float_col) e1, max(float_col) e2, sum(double_col) e3
from functional.alltypes) v
---- PLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(3,6,9,12,13) = 3, count(tinyint_col)), aggif(valid_tid(3,6,9,12,13) = 9, avg(int_col)), aggif(valid_tid(3,6,9,12,13) = 13, max(float_col))
| row-size=20B cardinality=1
|
02:AGGREGATE [FINALIZE]
| Class 0
| output: count(tinyint_col)
| Class 1
| output: avg(int_col)
| Class 2
| output: max:merge(float_col)
| row-size=20B cardinality=3
|
01:AGGREGATE
| Class 0
| group by: tinyint_col
| Class 1
| group by: int_col
| Class 2
| output: max(float_col)
| row-size=9B cardinality=21
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=9B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(3,6,9,12,13) = 3, count(tinyint_col)), aggif(valid_tid(3,6,9,12,13) = 9, avg(int_col)), aggif(valid_tid(3,6,9,12,13) = 13, max(float_col))
| row-size=20B cardinality=1
|
07:AGGREGATE [FINALIZE]
| Class 0
| output: count:merge(tinyint_col)
| Class 1
| output: avg:merge(int_col)
| Class 2
| output: max:merge(float_col)
| row-size=20B cardinality=3
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| Class 0
| output: count(tinyint_col)
| Class 1
| output: avg(int_col)
| Class 2
| output: max:merge(float_col)
| row-size=20B cardinality=3
|
05:AGGREGATE
| Class 0
| group by: tinyint_col
| Class 1
| group by: int_col
| Class 2
| output: max:merge(float_col)
| row-size=9B cardinality=21
|
04:EXCHANGE [HASH(CASE valid_tid(1,7,13) WHEN 1 THEN murmur_hash(tinyint_col) WHEN 7 THEN murmur_hash(int_col) WHEN 13 THEN 0 END)]
|
01:AGGREGATE [STREAMING]
| Class 0
| group by: tinyint_col
| Class 1
| group by: int_col
| Class 2
| output: max(float_col)
| row-size=9B cardinality=21
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=9B cardinality=7.30K
====
# A subset of aggregation classes are materialized. With group by.
select a, c, e1, e2, e3, gby2 from
(select count(distinct tinyint_col) a, avg(distinct smallint_col) b,
count(distinct int_col) c, avg(distinct bigint_col) d,
min(float_col) e1, max(float_col) e2, sum(double_col) e3,
string_col gby1, date_string_col gby2
from functional.alltypes
group by gby1, gby2) v
---- PLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,5,7,10,11) = 2, count(tinyint_col)), aggif(valid_tid(2,5,7,10,11) = 7, count(int_col)), aggif(valid_tid(2,5,7,10,11) = 11, min(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, max(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, sum(double_col))
| group by: CASE valid_tid(2,7,11) WHEN 2 THEN string_col WHEN 7 THEN string_col WHEN 11 THEN string_col END, CASE valid_tid(2,7,11) WHEN 2 THEN date_string_col WHEN 7 THEN date_string_col WHEN 11 THEN date_string_col END
| row-size=56B cardinality=8.11K
|
02:AGGREGATE [FINALIZE]
| Class 0
| output: count(tinyint_col)
| group by: string_col, date_string_col
| Class 1
| output: count(int_col)
| group by: string_col, date_string_col
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| group by: string_col, date_string_col
| row-size=131B cardinality=22.08K
|
01:AGGREGATE
| Class 0
| group by: string_col, date_string_col, tinyint_col
| Class 1
| group by: string_col, date_string_col, int_col
| Class 2
| output: min(float_col), max(float_col), sum(double_col)
| group by: string_col, date_string_col
| row-size=120B cardinality=21.90K
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=50B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
08:EXCHANGE [UNPARTITIONED]
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,5,7,10,11) = 2, count(tinyint_col)), aggif(valid_tid(2,5,7,10,11) = 7, count(int_col)), aggif(valid_tid(2,5,7,10,11) = 11, min(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, max(float_col)), aggif(valid_tid(2,5,7,10,11) = 11, sum(double_col))
| group by: CASE valid_tid(2,7,11) WHEN 2 THEN string_col WHEN 7 THEN string_col WHEN 11 THEN string_col END, CASE valid_tid(2,7,11) WHEN 2 THEN date_string_col WHEN 7 THEN date_string_col WHEN 11 THEN date_string_col END
| row-size=56B cardinality=8.11K
|
07:AGGREGATE [FINALIZE]
| Class 0
| output: count:merge(tinyint_col)
| group by: string_col, date_string_col
| Class 1
| output: count:merge(int_col)
| group by: string_col, date_string_col
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| group by: string_col, date_string_col
| row-size=131B cardinality=22.08K
|
06:EXCHANGE [HASH(CASE valid_tid(2,7,11) WHEN 2 THEN murmur_hash(string_col) WHEN 7 THEN murmur_hash(string_col) WHEN 11 THEN murmur_hash(string_col) END,CASE valid_tid(2,7,11) WHEN 2 THEN murmur_hash(date_string_col) WHEN 7 THEN murmur_hash(date_string_col) WHEN 11 THEN murmur_hash(date_string_col) END)]
|
02:AGGREGATE [STREAMING]
| Class 0
| output: count(tinyint_col)
| group by: string_col, date_string_col
| Class 1
| output: count(int_col)
| group by: string_col, date_string_col
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| group by: string_col, date_string_col
| row-size=131B cardinality=22.08K
|
05:AGGREGATE
| Class 0
| group by: string_col, date_string_col, tinyint_col
| Class 1
| group by: string_col, date_string_col, int_col
| Class 2
| output: min:merge(float_col), max:merge(float_col), sum:merge(double_col)
| group by: string_col, date_string_col
| row-size=120B cardinality=21.90K
|
04:EXCHANGE [HASH(CASE valid_tid(1,6,11) WHEN 1 THEN murmur_hash(string_col) WHEN 6 THEN murmur_hash(string_col) WHEN 11 THEN murmur_hash(string_col) END,CASE valid_tid(1,6,11) WHEN 1 THEN murmur_hash(date_string_col) WHEN 6 THEN murmur_hash(date_string_col) WHEN 11 THEN murmur_hash(date_string_col) END,CASE valid_tid(1,6,11) WHEN 1 THEN murmur_hash(tinyint_col) WHEN 6 THEN murmur_hash(int_col) WHEN 11 THEN 0 END)]
|
01:AGGREGATE [STREAMING]
| Class 0
| group by: string_col, date_string_col, tinyint_col
| Class 1
| group by: string_col, date_string_col, int_col
| Class 2
| output: min(float_col), max(float_col), sum(double_col)
| group by: string_col, date_string_col
| row-size=120B cardinality=21.90K
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=50B cardinality=7.30K
====
# A subset of aggregation classes are materialized. A subset of aggregation functions
# of the surviving classes are materialized. With group by.
select a1, c2, e2, gby1 from
(select count(distinct tinyint_col) a1, avg(distinct tinyint_col) a2,
count(distinct smallint_col) b1, avg(distinct smallint_col) b2,
count(distinct int_col) c1, avg(distinct int_col) c2,
count(distinct bigint_col) d1, avg(distinct bigint_col) d2,
min(float_col) e1, max(float_col) e2, sum(double_col) e3,
string_col gby1, date_string_col gby2
from functional.alltypes
group by gby1, gby2) v
---- PLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(3,6,9,12,13) = 3, count(tinyint_col)), aggif(valid_tid(3,6,9,12,13) = 9, avg(int_col)), aggif(valid_tid(3,6,9,12,13) = 13, max(float_col))
| group by: CASE valid_tid(3,9,13) WHEN 3 THEN string_col WHEN 9 THEN string_col WHEN 13 THEN string_col END, CASE valid_tid(3,9,13) WHEN 3 THEN date_string_col WHEN 9 THEN date_string_col WHEN 13 THEN date_string_col END
| row-size=44B cardinality=8.11K
|
02:AGGREGATE [FINALIZE]
| Class 0
| output: count(tinyint_col)
| group by: string_col, date_string_col
| Class 1
| output: avg(int_col)
| group by: string_col, date_string_col
| Class 2
| output: max:merge(float_col)
| group by: string_col, date_string_col
| row-size=119B cardinality=22.08K
|
01:AGGREGATE
| Class 0
| group by: string_col, date_string_col, tinyint_col
| Class 1
| group by: string_col, date_string_col, int_col
| Class 2
| output: max(float_col)
| group by: string_col, date_string_col
| row-size=108B cardinality=21.90K
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=42B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
08:EXCHANGE [UNPARTITIONED]
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(3,6,9,12,13) = 3, count(tinyint_col)), aggif(valid_tid(3,6,9,12,13) = 9, avg(int_col)), aggif(valid_tid(3,6,9,12,13) = 13, max(float_col))
| group by: CASE valid_tid(3,9,13) WHEN 3 THEN string_col WHEN 9 THEN string_col WHEN 13 THEN string_col END, CASE valid_tid(3,9,13) WHEN 3 THEN date_string_col WHEN 9 THEN date_string_col WHEN 13 THEN date_string_col END
| row-size=44B cardinality=8.11K
|
07:AGGREGATE [FINALIZE]
| Class 0
| output: count:merge(tinyint_col)
| group by: string_col, date_string_col
| Class 1
| output: avg:merge(int_col)
| group by: string_col, date_string_col
| Class 2
| output: max:merge(float_col)
| group by: string_col, date_string_col
| row-size=119B cardinality=22.08K
|
06:EXCHANGE [HASH(CASE valid_tid(2,8,13) WHEN 2 THEN murmur_hash(string_col) WHEN 8 THEN murmur_hash(string_col) WHEN 13 THEN murmur_hash(string_col) END,CASE valid_tid(2,8,13) WHEN 2 THEN murmur_hash(date_string_col) WHEN 8 THEN murmur_hash(date_string_col) WHEN 13 THEN murmur_hash(date_string_col) END)]
|
02:AGGREGATE [STREAMING]
| Class 0
| output: count(tinyint_col)
| group by: string_col, date_string_col
| Class 1
| output: avg(int_col)
| group by: string_col, date_string_col
| Class 2
| output: max:merge(float_col)
| group by: string_col, date_string_col
| row-size=119B cardinality=22.08K
|
05:AGGREGATE
| Class 0
| group by: string_col, date_string_col, tinyint_col
| Class 1
| group by: string_col, date_string_col, int_col
| Class 2
| output: max:merge(float_col)
| group by: string_col, date_string_col
| row-size=108B cardinality=21.90K
|
04:EXCHANGE [HASH(CASE valid_tid(1,7,13) WHEN 1 THEN murmur_hash(string_col) WHEN 7 THEN murmur_hash(string_col) WHEN 13 THEN murmur_hash(string_col) END,CASE valid_tid(1,7,13) WHEN 1 THEN murmur_hash(date_string_col) WHEN 7 THEN murmur_hash(date_string_col) WHEN 13 THEN murmur_hash(date_string_col) END,CASE valid_tid(1,7,13) WHEN 1 THEN murmur_hash(tinyint_col) WHEN 7 THEN murmur_hash(int_col) WHEN 13 THEN 0 END)]
|
01:AGGREGATE [STREAMING]
| Class 0
| group by: string_col, date_string_col, tinyint_col
| Class 1
| group by: string_col, date_string_col, int_col
| Class 2
| output: max(float_col)
| group by: string_col, date_string_col
| row-size=108B cardinality=21.90K
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=42B cardinality=7.30K
====
# Simplifies to a single aggrgeation class. Only first distinct agg is materialized.
select a from
(select count(distinct tinyint_col) a, count(distinct smallint_col) b
from functional.alltypes) v
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(tinyint_col)
| row-size=8B cardinality=1
|
01:AGGREGATE
| group by: tinyint_col
| row-size=1B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=1B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
07:AGGREGATE [FINALIZE]
| output: count:merge(tinyint_col)
| row-size=8B cardinality=1
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| output: count(tinyint_col)
| row-size=8B cardinality=1
|
05:AGGREGATE
| group by: tinyint_col
| row-size=1B cardinality=10
|
04:EXCHANGE [HASH(tinyint_col)]
|
01:AGGREGATE [STREAMING]
| group by: tinyint_col
| row-size=1B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=1B cardinality=7.30K
====
# Simplifies to a single aggrgeation class. Only second distinct agg is materialized.
select b from
(select count(distinct tinyint_col) a, count(distinct smallint_col) b
from functional.alltypes) v
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(smallint_col)
| row-size=8B cardinality=1
|
01:AGGREGATE
| group by: smallint_col
| row-size=2B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=2B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
07:AGGREGATE [FINALIZE]
| output: count:merge(smallint_col)
| row-size=8B cardinality=1
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| output: count(smallint_col)
| row-size=8B cardinality=1
|
05:AGGREGATE
| group by: smallint_col
| row-size=2B cardinality=10
|
04:EXCHANGE [HASH(smallint_col)]
|
01:AGGREGATE [STREAMING]
| group by: smallint_col
| row-size=2B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=2B cardinality=7.30K
====
# Some aggs only referenced in HAVING clause.
select count(distinct tinyint_col), min(timestamp_col) from functional.alltypes
having count(distinct smallint_col) < 10 and max(date_string_col) = 'test'
---- PLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,4,5) = 2, count(tinyint_col)), aggif(valid_tid(2,4,5) = 4, count(smallint_col)), aggif(valid_tid(2,4,5) = 5, min(timestamp_col)), aggif(valid_tid(2,4,5) = 5, max(date_string_col))
| having: aggif(valid_tid(2,4,5) = 4, count(smallint_col)) < 10, aggif(valid_tid(2,4,5) = 5, max(date_string_col)) = 'test'
| row-size=44B cardinality=0
|
02:AGGREGATE [FINALIZE]
| Class 0
| output: count(tinyint_col)
| Class 1
| output: count(smallint_col)
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| row-size=44B cardinality=3
|
01:AGGREGATE
| Class 0
| group by: tinyint_col
| Class 1
| group by: smallint_col
| Class 2
| output: min(timestamp_col), max(date_string_col)
| row-size=31B cardinality=21
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=39B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,4,5) = 2, count(tinyint_col)), aggif(valid_tid(2,4,5) = 4, count(smallint_col)), aggif(valid_tid(2,4,5) = 5, min(timestamp_col)), aggif(valid_tid(2,4,5) = 5, max(date_string_col))
| having: aggif(valid_tid(2,4,5) = 4, count(smallint_col)) < 10, aggif(valid_tid(2,4,5) = 5, max(date_string_col)) = 'test'
| row-size=44B cardinality=0
|
07:AGGREGATE [FINALIZE]
| Class 0
| output: count:merge(tinyint_col)
| Class 1
| output: count:merge(smallint_col)
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| row-size=44B cardinality=3
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| Class 0
| output: count(tinyint_col)
| Class 1
| output: count(smallint_col)
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| row-size=44B cardinality=3
|
05:AGGREGATE
| Class 0
| group by: tinyint_col
| Class 1
| group by: smallint_col
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| row-size=31B cardinality=21
|
04:EXCHANGE [HASH(CASE valid_tid(1,3,5) WHEN 1 THEN murmur_hash(tinyint_col) WHEN 3 THEN murmur_hash(smallint_col) WHEN 5 THEN 0 END)]
|
01:AGGREGATE [STREAMING]
| Class 0
| group by: tinyint_col
| Class 1
| group by: smallint_col
| Class 2
| output: min(timestamp_col), max(date_string_col)
| row-size=31B cardinality=21
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=39B cardinality=7.30K
====
# Some aggs only referenced in ORDER BY clause.
select count(distinct tinyint_col), min(timestamp_col) from functional.alltypes
group by bigint_col
order by count(distinct smallint_col), max(date_string_col)
---- PLAN
PLAN-ROOT SINK
|
04:SORT
| order by: aggif(valid_tid(2,4,5) = 4, count(smallint_col)) ASC, aggif(valid_tid(2,4,5) = 5, max(date_string_col)) ASC
| row-size=44B cardinality=11
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,4,5) = 2, count(tinyint_col)), aggif(valid_tid(2,4,5) = 4, count(smallint_col)), aggif(valid_tid(2,4,5) = 5, min(timestamp_col)), aggif(valid_tid(2,4,5) = 5, max(date_string_col))
| group by: CASE valid_tid(2,4,5) WHEN 2 THEN bigint_col WHEN 4 THEN bigint_col WHEN 5 THEN bigint_col END
| row-size=52B cardinality=11
|
02:AGGREGATE [FINALIZE]
| Class 0
| output: count(tinyint_col)
| group by: bigint_col
| Class 1
| output: count(smallint_col)
| group by: bigint_col
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| group by: bigint_col
| row-size=68B cardinality=30
|
01:AGGREGATE
| Class 0
| group by: bigint_col, tinyint_col
| Class 1
| group by: bigint_col, smallint_col
| Class 2
| output: min(timestamp_col), max(date_string_col)
| group by: bigint_col
| row-size=55B cardinality=210
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=47B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
09:MERGING-EXCHANGE [UNPARTITIONED]
| order by: aggif(valid_tid(2,4,5) = 4, count(smallint_col)) ASC, aggif(valid_tid(2,4,5) = 5, max(date_string_col)) ASC
|
04:SORT
| order by: aggif(valid_tid(2,4,5) = 4, count(smallint_col)) ASC, aggif(valid_tid(2,4,5) = 5, max(date_string_col)) ASC
| row-size=44B cardinality=11
|
03:AGGREGATE [FINALIZE]
| output: aggif(valid_tid(2,4,5) = 2, count(tinyint_col)), aggif(valid_tid(2,4,5) = 4, count(smallint_col)), aggif(valid_tid(2,4,5) = 5, min(timestamp_col)), aggif(valid_tid(2,4,5) = 5, max(date_string_col))
| group by: CASE valid_tid(2,4,5) WHEN 2 THEN bigint_col WHEN 4 THEN bigint_col WHEN 5 THEN bigint_col END
| row-size=52B cardinality=11
|
08:AGGREGATE [FINALIZE]
| Class 0
| output: count:merge(tinyint_col)
| group by: bigint_col
| Class 1
| output: count:merge(smallint_col)
| group by: bigint_col
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| group by: bigint_col
| row-size=68B cardinality=30
|
07:EXCHANGE [HASH(CASE valid_tid(2,4,5) WHEN 2 THEN murmur_hash(bigint_col) WHEN 4 THEN murmur_hash(bigint_col) WHEN 5 THEN murmur_hash(bigint_col) END)]
|
02:AGGREGATE [STREAMING]
| Class 0
| output: count(tinyint_col)
| group by: bigint_col
| Class 1
| output: count(smallint_col)
| group by: bigint_col
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| group by: bigint_col
| row-size=68B cardinality=30
|
06:AGGREGATE
| Class 0
| group by: bigint_col, tinyint_col
| Class 1
| group by: bigint_col, smallint_col
| Class 2
| output: min:merge(timestamp_col), max:merge(date_string_col)
| group by: bigint_col
| row-size=55B cardinality=210
|
05:EXCHANGE [HASH(CASE valid_tid(1,3,5) WHEN 1 THEN murmur_hash(bigint_col) WHEN 3 THEN murmur_hash(bigint_col) WHEN 5 THEN murmur_hash(bigint_col) END,CASE valid_tid(1,3,5) WHEN 1 THEN murmur_hash(tinyint_col) WHEN 3 THEN murmur_hash(smallint_col) WHEN 5 THEN 0 END)]
|
01:AGGREGATE [STREAMING]
| Class 0
| group by: bigint_col, tinyint_col
| Class 1
| group by: bigint_col, smallint_col
| Class 2
| output: min(timestamp_col), max(date_string_col)
| group by: bigint_col
| row-size=55B cardinality=210
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=47B cardinality=7.30K
====
# Mixed distinct and non-distinct aggs. No materialized aggregations. No group by.
select 1 from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d
from functional.alltypes) v
---- PLAN
PLAN-ROOT SINK
|
01:AGGREGATE [FINALIZE]
| row-size=0B cardinality=1
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
partition key scan
row-size=0B cardinality=24
====
# Mixed distinct and non-distinct aggs. No materialized aggregations. With group by.
select v.gby1 from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d,
string_col gby1, date_string_col gby2
from functional.alltypes
group by gby1, gby2) v
---- PLAN
PLAN-ROOT SINK
|
01:AGGREGATE [FINALIZE]
| group by: string_col, date_string_col
| row-size=33B cardinality=7.30K
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=33B cardinality=7.30K
====
# Only distinct aggs. No materialized aggregations. No group by.
select 1 from
(select count(distinct tinyint_col) a, count(distinct smallint_col) b
from functional.alltypes) v
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| row-size=0B cardinality=1
|
01:AGGREGATE
| group by: smallint_col
| row-size=2B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=2B cardinality=7.30K
====
# Only distinct aggs. No materialized aggregations. With group by.
select v.gby2 from
(select count(distinct tinyint_col) a, count(distinct smallint_col) b,
string_col gby1, date_string_col gby2
from functional.alltypes
group by gby1, gby2) v
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| group by: string_col, date_string_col
| row-size=33B cardinality=7.30K
|
01:AGGREGATE
| group by: string_col, date_string_col, smallint_col
| row-size=35B cardinality=7.30K
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=35B cardinality=7.30K
====
# Simplifies to a single aggregation class. Only first distinct agg is materialized.
# No group by.
select b from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d
from functional.alltypes
having count(distinct tinyint_col) < 9) v
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(tinyint_col)
| having: count(tinyint_col) < 9
| row-size=8B cardinality=0
|
01:AGGREGATE
| group by: tinyint_col
| row-size=1B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=1B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
07:AGGREGATE [FINALIZE]
| output: count:merge(tinyint_col)
| having: count(tinyint_col) < 9
| row-size=8B cardinality=1
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| output: count(tinyint_col)
| row-size=8B cardinality=0
|
05:AGGREGATE
| group by: tinyint_col
| row-size=1B cardinality=10
|
04:EXCHANGE [HASH(tinyint_col)]
|
01:AGGREGATE [STREAMING]
| group by: tinyint_col
| row-size=1B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=1B cardinality=7.30K
====
# Simplifies to a single aggregation class. Only second distinct agg is materialized.
# No group by.
select d from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d
from functional.alltypes
having count(distinct smallint_col) < 9) v
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(smallint_col)
| having: count(smallint_col) < 9
| row-size=8B cardinality=0
|
01:AGGREGATE
| group by: smallint_col
| row-size=2B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=2B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
07:AGGREGATE [FINALIZE]
| output: count:merge(smallint_col)
| having: count(smallint_col) < 9
| row-size=8B cardinality=1
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| output: count(smallint_col)
| row-size=8B cardinality=0
|
05:AGGREGATE
| group by: smallint_col
| row-size=2B cardinality=10
|
04:EXCHANGE [HASH(smallint_col)]
|
01:AGGREGATE [STREAMING]
| group by: smallint_col
| row-size=2B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=2B cardinality=7.30K
====
# Simplifies to a single aggregation class. Only non-distinct aggs remain.
# No group by.
select a, c from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d
from functional.alltypes
having min(string_col) < '9') v
where c > '0' and c < a
---- PLAN
PLAN-ROOT SINK
|
01:AGGREGATE [FINALIZE]
| output: min(string_col), max(string_col)
| having: max(string_col) > '0', min(string_col) < '9', max(string_col) < min(string_col)
| row-size=24B cardinality=0
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=13B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
04:AGGREGATE [FINALIZE]
| output: min:merge(string_col), max:merge(string_col)
| having: max(string_col) > '0', min(string_col) < '9', max(string_col) < min(string_col)
| row-size=24B cardinality=0
|
03:EXCHANGE [UNPARTITIONED]
|
01:AGGREGATE
| output: min(string_col), max(string_col)
| row-size=24B cardinality=1
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=13B cardinality=7.30K
====
# Simplifies to a single aggregation class with distinct and non-distinct aggs.
# No group by.
select a, b, c from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d
from functional.alltypes
having min(string_col) < '9' and count(distinct tinyint_col) = 10) v
where c > '0' and c < a
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(tinyint_col), min:merge(string_col), max:merge(string_col)
| having: count(tinyint_col) = 10, max(string_col) > '0', min(string_col) < '9', max(string_col) < min(string_col)
| row-size=32B cardinality=0
|
01:AGGREGATE
| output: min(string_col), max(string_col)
| group by: tinyint_col
| row-size=25B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=14B cardinality=7.30K
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
07:AGGREGATE [FINALIZE]
| output: count:merge(tinyint_col), min:merge(string_col), max:merge(string_col)
| having: count(tinyint_col) = 10, max(string_col) > '0', min(string_col) < '9', max(string_col) < min(string_col)
| row-size=32B cardinality=1
|
06:EXCHANGE [UNPARTITIONED]
|
02:AGGREGATE
| output: count(tinyint_col), min:merge(string_col), max:merge(string_col)
| row-size=32B cardinality=0
|
05:AGGREGATE
| output: min:merge(string_col), max:merge(string_col)
| group by: tinyint_col
| row-size=25B cardinality=10
|
04:EXCHANGE [HASH(tinyint_col)]
|
01:AGGREGATE [STREAMING]
| output: min(string_col), max(string_col)
| group by: tinyint_col
| row-size=25B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
row-size=14B cardinality=7.30K
====
# Simplifies to a single aggregation class with one distinct agg.
# With group by.
select b, gby1 from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d,
date_string_col gby1, timestamp_col gby2
from functional.alltypes
group by gby1, gby2
having count(distinct tinyint_col) < 10) v
where gby1 = 'test1'
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(tinyint_col)
| group by: date_string_col, timestamp_col
| having: count(tinyint_col) < 10
| row-size=44B cardinality=1
|
01:AGGREGATE
| group by: date_string_col, timestamp_col, tinyint_col
| row-size=37B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
predicates: functional.alltypes.date_string_col = 'test1'
row-size=37B cardinality=10
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
08:EXCHANGE [UNPARTITIONED]
|
07:AGGREGATE [FINALIZE]
| output: count:merge(tinyint_col)
| group by: date_string_col, timestamp_col
| having: count(tinyint_col) < 10
| row-size=44B cardinality=10
|
06:EXCHANGE [HASH(date_string_col,timestamp_col)]
|
02:AGGREGATE [STREAMING]
| output: count(tinyint_col)
| group by: date_string_col, timestamp_col
| row-size=44B cardinality=1
|
05:AGGREGATE
| group by: date_string_col, timestamp_col, tinyint_col
| row-size=37B cardinality=10
|
04:EXCHANGE [HASH(date_string_col,timestamp_col,tinyint_col)]
|
01:AGGREGATE [STREAMING]
| group by: date_string_col, timestamp_col, tinyint_col
| row-size=37B cardinality=10
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
predicates: functional.alltypes.date_string_col = 'test1'
row-size=37B cardinality=10
====
# Simplifies to a single aggregation class with two non-distinct aggs.
# With group by.
select a, c, gby1, gby2 from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d,
date_string_col gby1, timestamp_col gby2
from functional.alltypes
group by gby1, gby2
having count(distinct tinyint_col) < 10) v
where gby1 = 'test1' and gby1 < gby2
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(tinyint_col), min:merge(string_col), max:merge(string_col)
| group by: date_string_col, timestamp_col
| having: count(tinyint_col) < 10, date_string_col < timestamp_col
| row-size=68B cardinality=0
|
01:AGGREGATE
| output: min(string_col), max(string_col)
| group by: date_string_col, timestamp_col, tinyint_col
| row-size=61B cardinality=3
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
predicates: functional.alltypes.timestamp_col > NULL, functional.alltypes.date_string_col = 'test1'
row-size=50B cardinality=3
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
08:EXCHANGE [UNPARTITIONED]
|
07:AGGREGATE [FINALIZE]
| output: count:merge(tinyint_col), min:merge(string_col), max:merge(string_col)
| group by: date_string_col, timestamp_col
| having: count(tinyint_col) < 10, date_string_col < timestamp_col
| row-size=68B cardinality=3
|
06:EXCHANGE [HASH(date_string_col,timestamp_col)]
|
02:AGGREGATE [STREAMING]
| output: count(tinyint_col), min:merge(string_col), max:merge(string_col)
| group by: date_string_col, timestamp_col
| row-size=68B cardinality=0
|
05:AGGREGATE
| output: min:merge(string_col), max:merge(string_col)
| group by: date_string_col, timestamp_col, tinyint_col
| row-size=61B cardinality=3
|
04:EXCHANGE [HASH(date_string_col,timestamp_col,tinyint_col)]
|
01:AGGREGATE [STREAMING]
| output: min(string_col), max(string_col)
| group by: date_string_col, timestamp_col, tinyint_col
| row-size=61B cardinality=3
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
predicates: functional.alltypes.timestamp_col > NULL, functional.alltypes.date_string_col = 'test1'
row-size=50B cardinality=3
====
# Simplifies to a single aggregation class with one distinct and one non-distinct agg.
# With group by.
select c, d, gby1, gby2 from
(select min(string_col) a, count(distinct tinyint_col) b,
max(string_col) c, count(distinct smallint_col) d,
date_string_col gby1, timestamp_col gby2
from functional.alltypes
group by gby1, gby2
having count(distinct smallint_col) < 20) v
where gby1 = 'test1' and gby1 < gby2
---- PLAN
PLAN-ROOT SINK
|
02:AGGREGATE [FINALIZE]
| output: count(smallint_col), max:merge(string_col)
| group by: date_string_col, timestamp_col
| having: count(smallint_col) < 20, date_string_col < timestamp_col
| row-size=56B cardinality=0
|
01:AGGREGATE
| output: max(string_col)
| group by: date_string_col, timestamp_col, smallint_col
| row-size=50B cardinality=3
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
predicates: functional.alltypes.timestamp_col > NULL, functional.alltypes.date_string_col = 'test1'
row-size=51B cardinality=3
---- DISTRIBUTEDPLAN
PLAN-ROOT SINK
|
08:EXCHANGE [UNPARTITIONED]
|
07:AGGREGATE [FINALIZE]
| output: count:merge(smallint_col), max:merge(string_col)
| group by: date_string_col, timestamp_col
| having: count(smallint_col) < 20, date_string_col < timestamp_col
| row-size=56B cardinality=3
|
06:EXCHANGE [HASH(date_string_col,timestamp_col)]
|
02:AGGREGATE [STREAMING]
| output: count(smallint_col), max:merge(string_col)
| group by: date_string_col, timestamp_col
| row-size=56B cardinality=0
|
05:AGGREGATE
| output: max:merge(string_col)
| group by: date_string_col, timestamp_col, smallint_col
| row-size=50B cardinality=3
|
04:EXCHANGE [HASH(date_string_col,timestamp_col,smallint_col)]
|
01:AGGREGATE [STREAMING]
| output: max(string_col)
| group by: date_string_col, timestamp_col, smallint_col
| row-size=50B cardinality=3
|
00:SCAN HDFS [functional.alltypes]
HDFS partitions=24/24 files=24 size=478.45KB
predicates: functional.alltypes.timestamp_col > NULL, functional.alltypes.date_string_col = 'test1'
row-size=51B cardinality=3
====
Reference in New Issue View Git Blame Copy Permalink