mysql查询需要很长时间

Question

my below query is taking more then 10 seconds. How I can optimize this query. Please suggest which field to add index. This query is drupal view for distance search.

    SELECT DISTINCT location.lid AS location_lid, node.nid AS nid, location.name AS location_name, location_phone.phone AS location_phone_phone, location_fax.fax AS location_fax_fax, node.title AS node_title, node.language AS node_language, location.additional AS location_additional, location.city AS location_city, location.latitude AS location_latitude, location.longitude AS location_longitude, location.country AS location_country, location.postal_code AS location_postal_code, location.province AS location_province, location.street AS location_street, field_data_field_showroom_level.field_showroom_level_value AS field_data_field_showroom_level_field_showroom_level_value, field_data_field_showroom_type.field_showroom_type_value AS field_data_field_showroom_type_field_showroom_type_value, field_data_field_showroom_inventory_records.field_showroom_inventory_records_value AS field_data_field_showroom_inventory_records_field_showroom_i, field_data_field_record_type_id.field_record_type_id_value AS field_data_field_record_type_id_field_record_type_id_value, field_data_field_toto_gallery.field_toto_gallery_value AS field_data_field_toto_gallery_field_toto_gallery_value, 'node' AS field_data_field_showroom_location_node_entity_type, 'node' AS field_data_field_day_of_operation_node_entity_type, 'node' AS field_data_field_day_of_operation_comments_node_entity_type, 'node' AS field_data_field_showroom_open_node_entity_type, 'node' AS field_data_field_showroom_close_node_entity_type, 'node' AS field_data_field_appointment_needed_node_entity_type, (COALESCE(ACOS(0.83308162381476*COS(RADIANS(location.latitude))*(0.098658826854837*COS(RADIANS(location.longitude)) + -0.99512131716873*SIN(RADIANS(location.longitude))) + 0.55315007734083*SIN(RADIANS(location.latitude))), 0.00000)*6371570.9190939) AS location_distance, 'node' AS field_data_field_salesforce_id_node_entity_type, 'node' AS field_data_field_multiple_product_sku_node_entity_type, 'node' AS field_data_field_product_types_node_entity_type, 'node' AS field_data_field_toto_gallery_node_entity_type, 'node' AS field_data_field_showroom_level_node_entity_type
FROM 
node node
LEFT JOIN location_instance location_instance ON node.vid = location_instance.vid
LEFT JOIN location location ON location_instance.lid = location.lid
INNER JOIN field_data_field_toto_gallery field_data_field_toto_gallery ON node.nid = field_data_field_toto_gallery.entity_id AND (field_data_field_toto_gallery.entity_type = 'node' AND field_data_field_toto_gallery.deleted = '0')
LEFT JOIN field_data_field_multiple_product_sku field_data_field_multiple_product_sku ON node.nid = field_data_field_multiple_product_sku.entity_id AND (field_data_field_multiple_product_sku.entity_type = 'node' AND field_data_field_multiple_product_sku.deleted = '0')
LEFT JOIN location_phone location_phone ON location_instance.lid = location_phone.lid
LEFT JOIN location_fax location_fax ON location_instance.lid = location_fax.lid
LEFT JOIN field_data_field_showroom_level field_data_field_showroom_level ON node.nid = field_data_field_showroom_level.entity_id AND (field_data_field_showroom_level.entity_type = 'node' AND field_data_field_showroom_level.deleted = '0')
LEFT JOIN field_data_field_showroom_type field_data_field_showroom_type ON node.nid = field_data_field_showroom_type.entity_id AND (field_data_field_showroom_type.entity_type = 'node' AND field_data_field_showroom_type.deleted = '0')
LEFT JOIN field_data_field_showroom_inventory_records field_data_field_showroom_inventory_records ON node.nid = field_data_field_showroom_inventory_records.entity_id AND (field_data_field_showroom_inventory_records.entity_type = 'node' AND field_data_field_showroom_inventory_records.deleted = '0')
LEFT JOIN field_data_field_record_type_id field_data_field_record_type_id ON node.nid = field_data_field_record_type_id.entity_id AND (field_data_field_record_type_id.entity_type = 'node' AND field_data_field_record_type_id.deleted = '0')
WHERE ((    (node.status = '1') 
        AND (node.type IN  ('showrooms')) 
           AND (location.latitude > '32.859795378699'
           AND location.latitude < '34.306986221301'
           AND location.longitude > '-85.206632058424'
           AND location.longitude < '-83.469478341576')
        AND ((COALESCE(ACOS(0.83308162574562*COS(RADIANS(location.latitude))*(0.098658823381208*COS(RADIANS(location.longitude)) + -0.99512131751311*SIN(RADIANS(location.longitude))) + 0.55315007443282*SIN(RADIANS(location.latitude))), 0.00000)*6371570.9191628) < '80467.35')
        AND (location.province = 'GA') 
        AND (location.city LIKE 'Atlanta' ESCAPE '\\') )
        AND( (field_data_field_toto_gallery.field_toto_gallery_value = '1')
           OR (field_data_field_multiple_product_sku.field_multiple_product_sku_value NOT LIKE 'NULL' ESCAPE '\\') ))

ORDER BY field_data_field_showroom_level_field_showroom_level_value DESC, 
         field_data_field_showroom_type_field_showroom_type_value DESC, 
         location_distance ASC, 
         field_data_field_showroom_inventory_records_field_showroom_i DESC,
         field_data_field_record_type_id_field_record_type_id_value ASC,          
         field_data_field_toto_gallery_field_toto_gallery_value DESC
LIMIT 10 OFFSET 0

Answer 1

I would recommend, that you learn about the PROCEDURE ANALYSE function of MySQL: http://dev.mysql.com/doc/refman/5.0/en/procedure-analyse.html

It will tell you which joins might not have an index they can use. From only looking at your query but without knowing the structure or data of the tables, we can't give you a good adivce. Usually you should have an index on every column, you use in a JOIN and ORDER BY part of you query and you might have some indexes for the columns used in the WHERE clause.

Answer 2

Here are some things to try immediately to see if you can improve performance.

1. Try a compound index on the location table on (province, city, latitude, longitude, lid) . That compound index will, hopefully, accelerate the WHERE criteria computation for your locations.
2. Make sure your latitude and longitude columns have the FLOAT or DOUBLE data type.
3. Try a compound index on the node table on (status, type, vid) . That index should help locate the exact node records in your WHERE clause and accelerate your joins.
4. Eliminate the spherical geometry computation from your WHERE clause. You already have the bounding box range restriction, and you're already sorting by distance, so you shouldn't get too many extra locations this way. And, if you do, they shouldn't be too far away by a factor of more than 4/pi.
5. Can you get rid of the SELECT DISTINCT and just use SELECT ? Finding distinct values is a data-intensive operation.

For some helpful information on the distance computation and the bounding box optimization, please see this. http://www.plumislandmedia.net/mysql/haversine-mysql-nearest-loc/