I have 2 database tables:
Message
table that contains gps message MessageData
that contains message sensors values I join them to get sensor values between 2 dates.
The query returns results slowly with 99% of time seeking the clustered index.
How to improve this performance? Any advice on the index or the query side?
I run the following query for join: I filter with primary key > 3264353049 to prevent seek dates with rows from last years
SELECT
t.MessageId, t.DataSourceId, t.[Value],
DataSource.SourceNameId, DataSource.Name
FROM
[MessageData] t
INNER JOIN
[Message] m ON t.MessageId = m.MessageId
LEFT JOIN
DataSource ON t.DataSourceId = DataSource.DataSourceId
WHERE
t.MessageId > 3264353049
AND m.ObjectId = @objId
AND m.GpsTime BETWEEN @dtFrom AND @dtTo
AND m.Valid = 1;
I have these indexes:
ALTER TABLE [MessageData]
ADD CONSTRAINT [AnalogData_PK]
PRIMARY KEY CLUSTERED ([MessageId] ASC, [DataSourceId] ASC)
CREATE NONCLUSTERED INDEX [MessageData_DataSourceId_IDX]
ON [MessageData] ([DataSourceId] ASC)
GO
CREATE NONCLUSTERED INDEX [IX_gpstime_objectid]
ON [Message] ([GpsTime] ASC)
INCLUDE ([MessageId], [ObjectId])
GO
CREATE UNIQUE CLUSTERED INDEX [Message_Object_UK]
ON [Message] ([ObjectId] ASC, [GpsTime] ASC, [MessageId] ASC)
GO
Message table data:
MessageId ObjectId VectorAngle VectorSpeed Altitude GpsTime X Y VisibleSatelites
-------------------- ----------- ----------- ----------- ----------- ------------------------------ ---------------------- ---------------------- ----------------
9988600080 192 0 0 0 2017-07-19 00:03:20 0 0 0
9988600082 192 0 0 0 2017-07-19 00:08:20 0 0 0
9988600086 192 0 0 0 2017-07-19 00:13:20 0 0 0
9988600089 192 0 0 0 2017-07-19 00:18:20 0 0 0
9988600092 192 0 0 0 2017-07-19 00:23:20 0 0 0
and MessageData
table data:
MessageId DataSourceId Value SourceNameId Name
-------------------- ------------ ---------------------- ------------ ------------------------------
9988600080 6364 0 1 Engine
9988600080 6365 0 2 Digital Input Status 2
9988600080 325346 0 179 DOUT 1
9988600080 325347 0 180 DOUT 2
9988600080 334214 0 69 Bettary
9988600082 6364 0 1 Engine
9988600082 6365 0 2 Digital Input Status 2
9988600082 325346 0 179 DOUT 1
9988600082 325347 0 180 DOUT 2
9988600082 334214 0 69 Bettary
9988600086 6364 0 1 Engine
9988600086 6365 0 2 Digital Input Status 2
9988600086 325346 0 179 DOUT 1
9988600086 325347 0 180 DOUT 2
9988600086 334214 0 69 Bettary
9988600089 6364 0 1 Engine
9988600089 6365 0 2 Digital Input Status 2
9988600089 325346 0 179 DOUT 1
9988600089 325347 0 180 DOUT 2
9988600089 334214 0 69 Bettary
9988600092 6364 0 1 Engine
9988600092 6365 0 2 Digital Input Status 2
9988600092 325346 0 179 DOUT 1
9988600092 325347 0 180 DOUT 2
9988600092 334214 0 69 Bettary
Well as mentioned in the comment I would definitely update my stats. But after that I would look at this index:
CREATE UNIQUE CLUSTERED INDEX [Message_Object_UK]
ON [Message] ([ObjectId] ASC, [GpsTime] ASC, [MessageId] ASC)
The query is doing that nasty Nested Loop Join because it has to seek in and find the records with the specific messageID that is has to pull out of EACH record in the Message table index. Try changing it to:
CREATE UNIQUE CLUSTERED INDEX [Message_Object_UK]
ON [Message] ([MessageId] ASC, [ObjectId] ASC, [GpsTime] ASC )
This has a chance of getting you a hash or merge join
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.