db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dag H. Wanvik (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (DERBY-6093) dramatically worse query plan for prepared stamement vs calling directly
Date Mon, 04 Mar 2013 06:05:12 GMT

    [ https://issues.apache.org/jira/browse/DERBY-6093?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13591683#comment-13591683
] 

Dag H. Wanvik edited comment on DERBY-6093 at 3/4/13 6:03 AM:
--------------------------------------------------------------

Does it make a difference if, in the prepared statement, you add "fe1.arg_0_term = fe2.arg_0_term"
? I see that the constants are the same in the direct SQL, so Derby seems to have used this
fact to avoid accessing the inner tables more than once: in the prepared statement plan it
still uses the index KB_FORMULA_ENTRIES_FORMULA_TERM_ARG0_NUM_TYPE but opens the inner table
3863 times - not realizing the two arg_0_term arguments are equal, it seems..

Update: Sorry, I see now they the good and the bad use different indexes:
good: KB_FORMULA_ENTRIES_FORMULA_TERM_TYPE                    and KB_FOMULA_ENTRIES_FORMULA_ID_FK
bad : KB_FORMULA_ENTRIES_FORMULA_TERM_ARG0_NUM_TYPE and KB_FORMULA_ENTRIES_FORMULA_TERM_ARG0_NUM_TYPE

Still, would adding the extra predicate make the ps version use the "good" indexes? (Not saying
you should do this in production, just curious if that's what affects the optimizer's different
choice.)

                
      was (Author: dagw):
    Does it make a difference if, in the prepared statement, you add "fe1.arg_0_term = fe2.arg_0_term"
? I see that the constants are the same in the direct SQL, so Derby seems to have used this
fact to avoid accessing the inner tables more than once: in the prepared statement plan it
still uses the index KB_FORMULA_ENTRIES_FORMULA_TERM_ARG0_NUM_TYPE but opens the inner table
3863 times - not realizing the two arg_0_term arguments are equal, it seems..
                  
> dramatically worse query plan for prepared stamement vs calling directly
> ------------------------------------------------------------------------
>
>                 Key: DERBY-6093
>                 URL: https://issues.apache.org/jira/browse/DERBY-6093
>             Project: Derby
>          Issue Type: Bug
>          Components: Store
>    Affects Versions: 10.9.1.0
>         Environment: Linux
>            Reporter: Tony Brusseau
>
> If I run a count query given all the parameters directly, then the query executes in
.05s, however, if I send it across as a prepared statement, then the query takes 40s to run
(because is does a complete table scan on a table with 15m rows).
> The query looks like:
> SELECT COUNT(*) 
> FROM KB.FORMULA_ENTRIES fe1, KB.FORMULA_ENTRIES fe2  
> WHERE 
> (fe1.arg_0_term = 1407374883620920) AND (fe1.arg_num = 1) AND (fe1.arg_term = 1407374883574780)
AND (fe1.formula_type = 0) 
> AND (fe1.formula_id = fe2.formula_id) AND (fe2.arg_0_term = 1407374883620920) AND (fe2.arg_num
= 2) AND (fe2.arg_term = 1407374883663337) AND (fe2.formula_type = 0)
>  
> or the prepared version like:
>  SELECT COUNT(*) 
> FROM KB.FORMULA_ENTRIES fe1 
> , KB.FORMULA_ENTRIES fe2  
> WHERE 
> (fe1.arg_0_term = ?) AND (fe1.arg_num = ?) AND (fe1.arg_term = ?) AND (fe1.formula_type
= ?) 
> AND (fe1.formula_id = fe2.formula_id) AND (fe2.arg_0_term = ?) AND (fe2.arg_num = ?)
AND (fe2.arg_term = ?) AND (fe2.formula_type = ?)
> The table is defined as:
> DROP TABLE KB.FORMULA_ENTRIES;
> CREATE TABLE KB.FORMULA_ENTRIES
> (
>         formula_entries_id           BIGINT NOT NULL,)
>         formula_id                   BIGINT NOT NULL,
>         arg_0_term                   BIGINT NOT NULL,
>         arg_term                     BIGINT NOT NULL,
>         arg_num                      INTEGER NOT NULL,
>         formula_type                 SMALLINT NOT NULL
> );
> ALTER TABLE kb.formula_entries ADD CONSTRAINT kb_formula_entries_pk PRIMARY KEY (formula_entries_id);
> ALTER TABLE kb.formula_entries ADD CONSTRAINT kb_fomula_entries_formula_id_fk 
>         FOREIGN KEY (formula_id) REFERENCES kb.formula_term (term_id) ON DELETE CASCADE;
> CREATE INDEX kb_formula_entries_formula_term_type ON kb.formula_entries (arg_term, formula_type);
> CREATE INDEX kb_formula_entries_formula_term_arg0_type ON kb.formula_entries (arg_term,
arg_0_term, formula_type);
> CREATE INDEX kb_formula_entries_formula_term_num_type ON kb.formula_entries (arg_term,
arg_num, formula_type);
> CREATE INDEX kb_formula_entries_formula_term_arg0_num_type ON kb.formula_entries (arg_term,
arg_0_term, arg_num, formula_type);
> *************************************************************************
> The good query plan (when *not* sent as a prepared statements) looks like:
> Tue Feb 26 16:15:21 CST 2013 Thread[DRDAConnThread_5,5,main] (XID = 40971291), (SESSIONID
= 7), SELECT COUNT(*) 
> FROM KB.FORMULA_ENTRIES fe1, KB.FORMULA_ENTRIES fe2  
> WHERE 
> (fe1.arg_0_term = 1407374883620920) AND (fe1.arg_num = 1) AND (fe1.arg_term = 1407374883574780)
AND (fe1.formula_type = 0) 
> AND (fe1.formula_id = fe2.formula_id) AND (fe2.arg_0_term = 1407374883620920) AND (fe2.arg_num
= 2) AND (fe2.arg_term = 1407374883663337) AND (fe2.formula_type = 0) ******* Project-Restrict
ResultSet (10):
> Number of opens = 1
> Rows seen = 1
> Rows filtered = 0
> restriction = false
> projection = true
> 	constructor time (milliseconds) = 0
> 	open time (milliseconds) = 0
> 	next time (milliseconds) = 0
> 	close time (milliseconds) = 0
> 	restriction time (milliseconds) = 0
> 	projection time (milliseconds) = 0
> 	optimizer estimated row count: 1.00
> 	optimizer estimated cost: 7184.99
> Source result set:
> 	Scalar Aggregate ResultSet:
> 	Number of opens = 1
> 	Rows input = 3863
> 		constructor time (milliseconds) = 0
> 		open time (milliseconds) = 0
> 		next time (milliseconds) = 0
> 		close time (milliseconds) = 0
> 		optimizer estimated row count: 0.00
> 		optimizer estimated cost: 7184.99
> 	Index Key Optimization = false
> 	Source result set:
> 		Project-Restrict ResultSet (9):
> 		Number of opens = 1
> 		Rows seen = 3863
> 		Rows filtered = 0
> 		restriction = false
> 		projection = true
> 			constructor time (milliseconds) = 0
> 			open time (milliseconds) = 0
> 			next time (milliseconds) = 0
> 			close time (milliseconds) = 0
> 			restriction time (milliseconds) = 0
> 			projection time (milliseconds) = 0
> 			optimizer estimated row count: 0.00
> 			optimizer estimated cost: 7184.99
> 		Source result set:
> 			Nested Loop Join ResultSet:
> 			Number of opens = 1
> 			Rows seen from the left = 3863
> 			Rows seen from the right = 3863
> 			Rows filtered = 0
> 			Rows returned = 3863
> 				constructor time (milliseconds) = 0
> 				open time (milliseconds) = 0
> 				next time (milliseconds) = 0
> 				close time (milliseconds) = 0
> 				optimizer estimated row count: 0.00
> 				optimizer estimated cost: 7184.99
> 			Left result set:
> 				Project-Restrict ResultSet (5):
> 				Number of opens = 1
> 				Rows seen = 3865
> 				Rows filtered = 2
> 				restriction = true
> 				projection = true
> 					constructor time (milliseconds) = 0
> 					open time (milliseconds) = 0
> 					next time (milliseconds) = 0
> 					close time (milliseconds) = 0
> 					restriction time (milliseconds) = 0
> 					projection time (milliseconds) = 0
> 					optimizer estimated row count: 35.89
> 					optimizer estimated cost: 6785.98
> 				Source result set:
> 					Index Row to Base Row ResultSet for FORMULA_ENTRIES:
> 					Number of opens = 1
> 					Rows seen = 3865
> 					Columns accessed from heap = {1, 2, 3, 4, 5}
> 						constructor time (milliseconds) = 0
> 						open time (milliseconds) = 0
> 						next time (milliseconds) = 0
> 						close time (milliseconds) = 0
> 						optimizer estimated row count: 35.89
> 						optimizer estimated cost: 6785.98
> 						Index Scan ResultSet for FORMULA_ENTRIES using index KB_FORMULA_ENTRIES_FORMULA_TERM_TYPE
at read committed isolation level using instantaneous share row locking chosen by the optimizer
> 						Number of opens = 1
> 						Rows seen = 3865
> 						Rows filtered = 0
> 						Fetch Size = 16
> 							constructor time (milliseconds) = 0
> 							open time (milliseconds) = 0
> 							next time (milliseconds) = 0
> 							close time (milliseconds) = 0
> 							next time in milliseconds/row = 0
> 						scan information:
> 							Bit set of columns fetched=All
> 							Number of columns fetched=3
> 							Number of deleted rows visited=0
> 							Number of pages visited=7
> 							Number of rows qualified=3865
> 							Number of rows visited=3866
> 							Scan type=btree
> 							Tree height=3
> 							start position:
> 								>= on first 2 column(s).
> 								Ordered null semantics on the following columns: 
> 								0 1 
> 							stop position:
> 								> on first 2 column(s).
> 								Ordered null semantics on the following columns: 
> 								0 1 
> 							qualifiers:
> 								None
> 							optimizer estimated row count: 35.89
> 							optimizer estimated cost: 6785.98
> 			Right result set:
> 				Project-Restrict ResultSet (8):
> 				Number of opens = 3863
> 				Rows seen = 15452
> 				Rows filtered = 11589
> 				restriction = true
> 				projection = true
> 					constructor time (milliseconds) = 0
> 					open time (milliseconds) = 0
> 					next time (milliseconds) = 0
> 					close time (milliseconds) = 0
> 					restriction time (milliseconds) = 0
> 					projection time (milliseconds) = 0
> 					optimizer estimated row count: 0.00
> 					optimizer estimated cost: 399.01
> 				Source result set:
> 					Index Row to Base Row ResultSet for FORMULA_ENTRIES:
> 					Number of opens = 3863
> 					Rows seen = 15452
> 					Columns accessed from heap = {1, 2, 3, 4, 5}
> 						constructor time (milliseconds) = 0
> 						open time (milliseconds) = 0
> 						next time (milliseconds) = 0
> 						close time (milliseconds) = 0
> 						optimizer estimated row count: 0.00
> 						optimizer estimated cost: 399.01
> 						Index Scan ResultSet for FORMULA_ENTRIES using constraint KB_FOMULA_ENTRIES_FORMULA_ID_FK
at read committed isolation level using instantaneous share row locking chosen by the optimizer
> 						Number of opens = 3863
> 						Rows seen = 15452
> 						Rows filtered = 0
> 						Fetch Size = 16
> 							constructor time (milliseconds) = 0
> 							open time (milliseconds) = 0
> 							next time (milliseconds) = 0
> 							close time (milliseconds) = 0
> 							next time in milliseconds/row = 0
> 						scan information:
> 							Bit set of columns fetched=All
> 							Number of columns fetched=2
> 							Number of deleted rows visited=0
> 							Number of pages visited=11606
> 							Number of rows qualified=15452
> 							Number of rows visited=19315
> 							Scan type=btree
> 							Tree height=3
> 							start position:
> 								>= on first 1 column(s).
> 								Ordered null semantics on the following columns: 
> 								0 
> 							stop position:
> 								> on first 1 column(s).
> 								Ordered null semantics on the following columns: 
> 								0 
> 							qualifiers:
> 								None
> 							optimizer estimated row count: 0.00
> 							optimizer estimated cost: 399.01
> *************************************************************************
> The bad query plan (when sent as a prepared statements) looks like:
> Tue Feb 26 16:09:35 CST 2013 Thread[DRDAConnThread_4,5,main] (XID = 40971265), (SESSIONID
= 5), SELECT COUNT(*) 
> FROM KB.FORMULA_ENTRIES fe1 
> , KB.FORMULA_ENTRIES fe2  
> WHERE 
> (fe1.arg_0_term = ?) AND (fe1.arg_num = ?) AND (fe1.arg_term = ?) AND (fe1.formula_type
= ?) 
> AND (fe1.formula_id = fe2.formula_id) AND (fe2.arg_0_term = ?) AND (fe2.arg_num = ?)
AND (fe2.arg_term = ?) AND (fe2.formula_type = ?) ******* Project-Restrict ResultSet (9):
> Number of opens = 1
> Rows seen = 1
> Rows filtered = 0
> restriction = false
> projection = true
> 	constructor time (milliseconds) = 0
> 	open time (milliseconds) = 0
> 	next time (milliseconds) = 0
> 	close time (milliseconds) = 0
> 	restriction time (milliseconds) = 0
> 	projection time (milliseconds) = 0
> 	optimizer estimated row count: 1.00
> 	optimizer estimated cost: 43.36
> Source result set:
> 	Scalar Aggregate ResultSet:
> 	Number of opens = 1
> 	Rows input = 3863
> 		constructor time (milliseconds) = 0
> 		open time (milliseconds) = 0
> 		next time (milliseconds) = 0
> 		close time (milliseconds) = 0
> 		optimizer estimated row count: 0.00
> 		optimizer estimated cost: 43.36
> 	Index Key Optimization = false
> 	Source result set:
> 		Project-Restrict ResultSet (8):
> 		Number of opens = 1
> 		Rows seen = 3863
> 		Rows filtered = 0
> 		restriction = false
> 		projection = true
> 			constructor time (milliseconds) = 0
> 			open time (milliseconds) = 0
> 			next time (milliseconds) = 0
> 			close time (milliseconds) = 0
> 			restriction time (milliseconds) = 0
> 			projection time (milliseconds) = 0
> 			optimizer estimated row count: 0.00
> 			optimizer estimated cost: 43.36
> 		Source result set:
> 			Nested Loop Join ResultSet:
> 			Number of opens = 1
> 			Rows seen from the left = 3863
> 			Rows seen from the right = 3863
> 			Rows filtered = 0
> 			Rows returned = 3863
> 				constructor time (milliseconds) = 0
> 				open time (milliseconds) = 0
> 				next time (milliseconds) = 0
> 				close time (milliseconds) = 0
> 				optimizer estimated row count: 0.00
> 				optimizer estimated cost: 43.36
> 			Left result set:
> 				Index Row to Base Row ResultSet for FORMULA_ENTRIES:
> 				Number of opens = 1
> 				Rows seen = 3863
> 				Columns accessed from heap = {1, 2, 3, 4, 5}
> 					constructor time (milliseconds) = 0
> 					open time (milliseconds) = 0
> 					next time (milliseconds) = 0
> 					close time (milliseconds) = 0
> 					optimizer estimated row count: 2.98
> 					optimizer estimated cost: 10.89
> 					Index Scan ResultSet for FORMULA_ENTRIES using index KB_FORMULA_ENTRIES_FORMULA_TERM_ARG0_NUM_TYPE
at read committed isolation level using instantaneous share row locking chosen by the optimizer
> 					Number of opens = 1
> 					Rows seen = 3863
> 					Rows filtered = 0
> 					Fetch Size = 16
> 						constructor time (milliseconds) = 0
> 						open time (milliseconds) = 0
> 						next time (milliseconds) = 0
> 						close time (milliseconds) = 0
> 						next time in milliseconds/row = 0
> 					scan information:
> 						Bit set of columns fetched=All
> 						Number of columns fetched=5
> 						Number of deleted rows visited=0
> 						Number of pages visited=8
> 						Number of rows qualified=3863
> 						Number of rows visited=3864
> 						Scan type=btree
> 						Tree height=3
> 						start position:
> 							>= on first 4 column(s).
> 							Ordered null semantics on the following columns: 
> 						stop position:
> 							> on first 4 column(s).
> 							Ordered null semantics on the following columns: 
> 						qualifiers:
> 							None
> 						optimizer estimated row count: 2.98
> 						optimizer estimated cost: 10.89
> 			Right result set:
> 				Project-Restrict ResultSet (7):
> 				Number of opens = 3863
> 				Rows seen = 14922769
> 				Rows filtered = 14918906
> 				restriction = true
> 				projection = true
> 					constructor time (milliseconds) = 0
> 					open time (milliseconds) = 0
> 					next time (milliseconds) = 0
> 					close time (milliseconds) = 0
> 					restriction time (milliseconds) = 0
> 					projection time (milliseconds) = 0
> 					optimizer estimated row count: 0.00
> 					optimizer estimated cost: 32.47
> 				Source result set:
> 					Index Row to Base Row ResultSet for FORMULA_ENTRIES:
> 					Number of opens = 3863
> 					Rows seen = 14922769
> 					Columns accessed from heap = {1, 2, 3, 4, 5}
> 						constructor time (milliseconds) = 0
> 						open time (milliseconds) = 0
> 						next time (milliseconds) = 0
> 						close time (milliseconds) = 0
> 						optimizer estimated row count: 0.00
> 						optimizer estimated cost: 32.47
> 						Index Scan ResultSet for FORMULA_ENTRIES using index KB_FORMULA_ENTRIES_FORMULA_TERM_ARG0_NUM_TYPE
at read committed isolation level using instantaneous share row locking chosen by the optimizer
> 						Number of opens = 3863
> 						Rows seen = 14922769
> 						Rows filtered = 0
> 						Fetch Size = 16
> 							constructor time (milliseconds) = 0
> 							open time (milliseconds) = 0
> 							next time (milliseconds) = 0
> 							close time (milliseconds) = 0
> 							next time in milliseconds/row = 0
> 						scan information:
> 							Bit set of columns fetched=All
> 							Number of columns fetched=5
> 							Number of deleted rows visited=0
> 							Number of pages visited=34767
> 							Number of rows qualified=14922769
> 							Number of rows visited=14926632
> 							Scan type=btree
> 							Tree height=3
> 							start position:
> 								>= on first 4 column(s).
> 								Ordered null semantics on the following columns: 
> 							stop position:
> 								> on first 4 column(s).
> 								Ordered null semantics on the following columns: 
> 							qualifiers:
> 								None
> 							optimizer estimated row count: 0.00
> 							optimizer estimated cost: 32.47

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message