[DRILL-7121] TPCH 4 takes longer when Statistics is disabled.

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 1.16.0
• Fix Version/s: 1.16.0
• Component/s: Query Planning & Optimization
• Assignee: Gautam Parai
• Reporter: Robert Hou
• Created: 19/Mar/19 23:14
• Updated: 28/Mar/19 23:28
• Resolved: 28/Mar/19 23:14

Description

Here is TPCH 4 with sf 100:

select
  o.o_orderpriority,
  count(*) as order_count
from
  orders o

where
  o.o_orderdate >= date '1996-10-01'
  and o.o_orderdate < date '1996-10-01' + interval '3' month
  and 
  exists (
    select
      *
    from
      lineitem l
    where
      l.l_orderkey = o.o_orderkey
      and l.l_commitdate < l.l_receiptdate
  )
group by
  o.o_orderpriority
order by
  o.o_orderpriority;

The plan has changed when Statistics is disabled. A Hash Agg and a Broadcast Exchange have been added. These two operators expand the number of rows from the lineitem table from 137M to 9B rows. This forces the hash join to use 6GB of memory instead of 30 MB.

Issue Links

• is duplicated by: DRILL-7120 Query fails with ChannelClosedException when Statistics is disabled
• is duplicated by: DRILL-7122 TPCDS queries 29 25 17 are slower when Statistics is disabled.
• is duplicated by: DRILL-7123 TPCDS query 83 runs slower when Statistics is disabled
• links to: GitHub Pull Request #1718