Why is my simple SQL statement taking so long to execute and how do i go about finding the issue? - sql

Question

I have a very simple SQL query:
select o.Visit_ID
from Datamart.dbo.ww_Orders o
inner join Datamart.dbo.ww_Order_Details on o.Visit_ID = ww_Order_Details.Visit_ID
where o.runstamp = '20160422'
this query takes < 0 seconds to return 11173 rows
When I add the GROUP BY statement:
select o.Visit_ID
from Datamart.dbo.ww_Orders o
inner join Datamart.dbo.ww_Order_Details on o.Visit_ID = ww_Order_Details.Visit_ID
where o.runstamp = '20160422'
group by o.Visit_ID
the server takes 6min 30 sec to retrieve the 3047 rows.
I would expect the GROUP BY query to take not that much longer than the original. How do I go about finding what the issues are? thanks
here are the table definitions:
Orders:
CREATE TABLE [dbo].[ww_Orders](
[Visit_ID] [int] NOT NULL,
[Member_ID] [int] NOT NULL,
[Membership_no] [varchar](20) NULL,
[Member_Card_Num_Orig] [varchar](16) NULL,
[SCV_ID] [int] NULL,
[Meeting_No] [int] NULL,
[Location_Name] [varchar](128) NULL,
[Leader_No] [int] NULL,
[CashAmt] [decimal](18, 2) NULL,
[EFTAmt] [decimal](18, 2) NULL,
[VouchAmt] [decimal](18, 2) NULL,
[Meet_Date] [datetime] NULL,
[runstamp] [varchar](50) NULL,
CONSTRAINT [PK_dbo.ww_Orders] PRIMARY KEY CLUSTERED
(
[Visit_ID] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]
GO
SET ANSI_PADDING OFF
GO
Order Details:
CREATE TABLE [dbo].[ww_Order_Details](
[ord_det_pk] [int] IDENTITY(1,1) NOT NULL,
[Visit_ID] [int] NOT NULL,
[Item_Code] [nvarchar](50) NULL,
[Item_Name] [nvarchar](50) NULL,
[Qty] [int] NULL,
[Amt] [decimal](18, 2) NULL,
[Category_Code] [nvarchar](20) NULL,
CONSTRAINT [PK_ww_Order_Details] PRIMARY KEY CLUSTERED
(
[ord_det_pk] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]
GO
ALTER TABLE [dbo].[ww_Order_Details] WITH CHECK ADD CONSTRAINT [FK_ww_Order_Details_ww_Orders] FOREIGN KEY([Visit_ID])
REFERENCES [dbo].[ww_Orders] ([Visit_ID])
GO
ALTER TABLE [dbo].[ww_Order_Details] CHECK CONSTRAINT [FK_ww_Order_Details_ww_Orders]
GO

Answer 1

I'd add an index onto ww_Order_Details Visit_ID, probably make it the clustered index and drop the index on ord_det_pk. Also it might make more sense as an exists?
select o.Visit_ID
from Datamart.dbo.ww_Orders o
where exists (select 0 from Datamart.dbo.ww_Order_Details where o.Visit_ID = ww_Order_Details.Visit_ID)
and o.runstamp = '20160422'

Answer 2

I usually try to use a common table expression in cases like these, when a part of the query is really fast but an addition of a simple operation makes it really slow, usually it helps.
Try:
WITH CTE AS (
select o.Visit_ID
from Datamart.dbo.ww_Orders o
inner join Datamart.dbo.ww_Order_Details od on o.Visit_ID = od.Visit_ID
where o.runstamp = '20160422'
)
SELECT Visit_ID FROM CTE
group by Visit_ID
If this helps you can try to compare execution plans for your original query and this version to see whats going on