如何加快此SQL查询的速度？

Question

I'm fairly new to SQL, and I'm trying to work out how to speed up a complex SQL query in postgres, perhaps by improving my use of indexes. This is the query:

SELECT
    (SELECT ev.code FROM classification_item ci, enumeration_value ev
        WHERE ev.key_id = :ak_0
        AND ci.entry_id = t.id AND ci.value_id = ev.id) AS axis_0,
    (SELECT ev.code FROM classification_item ci, enumeration_value ev
        WHERE ev.key_id = :ak_1
        AND ci.entry_id = t.id AND ci.value_id = ev.id) AS axis_1,
    SUM(t.amount) as amount,
    (SELECT ev.code FROM classification_item ci, enumeration_value ev
        WHERE ev.key_id = :key_time_id
        AND ci.entry_id = t.id AND ci.value_id = ev.id) AS time
FROM "entry" t
WHERE t.dataset_id = :dataset_id
AND t.id IN (SELECT ci.entry_id FROM classification_item ci, enumeration_value ev
    WHERE ev.key_id = :k_0
    AND ev.code = :v_0 AND ci.value_id = ev.id)
GROUP BY time, axis_0, axis_1

This is basically the database schema (as defined in Pylons):

table_dataset = Table('dataset', meta.metadata,
    Column('id', Integer, primary_key=True),
    )
table_entry = Table('entry', meta.metadata,
    Column('id', Integer, primary_key=True),
    Column('dataset_id', Integer, ForeignKey('dataset.id')),
    Column('amount', Float()),
    )
table_classification_item = Table('classification_item', meta.metadata,
    Column('id', Integer, primary_key=True),
    Column('entry_id', Integer, ForeignKey('entry.id'), index=True),
    Column('value_id', Integer, ForeignKey('enumeration_value.id'), index=True)
)
table_enumeration_value = Table('enumeration_value', meta.metadata,
    Column('id', Integer, primary_key=True),
    Column('key_id', Integer, ForeignKey('key.id'), index=True),
    Column('code', UnicodeText(), index=True),
    )

And it has indexes as follows:

"dataset_pkey" PRIMARY KEY, btree (id)
"entry_pkey" PRIMARY KEY, btree (id)
"classification_item_pkey" PRIMARY KEY, btree (id)
"ix_classification_item_entry_id" btree (entry_id)
"ix_classification_item_value_id" btree (value_id)
"enumeration_value_pkey" PRIMARY KEY, btree (id)
"ix_enumeration_value_code" btree (code)
"ix_enumeration_value_key_id" btree (key_id)

Am I missing any obvious index that would speed up the query? In particular:

• should I be using 'clustered' indexes?
• should I also be indexing amount on entry , or would that make no difference to SUM(t.amount) as amount ?

Thanks for your help. I know this is a pretty complex question, so please tell me if I can do anything to improve it.

------ UPDATE --------------

The output from EXPLAIN ANALYZE on the above query.

Answer 1

If the enumeration_value table is small, I guess you can get some improvment by making axis_1 and axis_0 as join and add an extra index.

something like this (not tested)

CREATE INDEX idx_ci_vi_ei ON classification_item(value_id, entry_id);

CREATE INDEX idx_id_ki ON enumeration_value(id, key_id);

SELECT
    ci_0.code AS axis_0,
    ci_1.code AS axis_1,
    SUM(t.amount) as amount,
    ci_t.code AS time
FROM 
   "entry" t,
   (SELECT ev.code FROM classification_item ci, enumeration_value ev
        WHERE ev.key_id = :ak_0 AND ci.value_id = ev.id) ci_0,
   (SELECT ev.code FROM classification_item ci, enumeration_value ev
        WHERE ev.key_id = :ak_1 AND ci.value_id = ev.id) ci_1,
   (SELECT ev.code FROM classification_item ci, enumeration_value ev
        WHERE ev.key_id = :key_time_id AND ci.value_id = ev.id) ci_t
WHERE t.dataset_id = :dataset_id 
AND t.id IN (SELECT ci.entry_id FROM classification_item ci, enumeration_value ev
    WHERE ev.key_id = :k_0
    AND ev.code = :v_0 AND ci.value_id = ev.id)
AND t.id = ci_0.entry_id AND t.id = ci_1.entry_id AND t.id = ci_t.entry_id
GROUP BY time, axis_0, axis_1

Answer 2

EXPLAIN ANALYZE对查询计划有何评价？