I am new to SSIS. I have data coming from a single source. I need to enter that data into several tables (these tables are related by foreign key relationships). I am usi
Staging table
CREATE TABLE [dbo].[Stage](
[Name] [varchar](50) NULL,
[Street] [varchar](50) NULL,
[State] [varchar](50) NULL,
[pkAddressID] [int] NULL
) ON [PRIMARY]
GO
Address table
CREATE TABLE [dbo].[Address](
[pkAddressID] [int] IDENTITY(1,1) NOT NULL,
[street] [varchar](50) NULL,
[state] [varchar](50) NULL
) ON [PRIMARY]
-- Employee table
CREATE TABLE [dbo].[Employee](
[pkEmployeeID] [int] IDENTITY(1,1) NOT NULL,
[fkAddressID] [int] NULL,
[Name] [varchar](50) NULL
) ON [PRIMARY]
GO
--- DFT - Load data into Stage table
--- Execute SQL Task 1 --- Populate Address table
Merge [dbo].[Address] as target
using
(
select distinct [Street], [State] from [dbo].[Stage]
) as source
on source.[Street] = target.[Street] and source.[State] = target.[State]
when not matched then
insert ([Street], [State])
values (source.[Street], source.[State])
;
--- Execute SQL Task 2 --- Populate Stage table//pkAddressID column
Merge [dbo].[Stage] as target
using
(
select [pkAddressID],[Street], [State] from [1Address]
)
as source
on source.[Street] = target.[Street] and source.[State] = target.[State]
when matched then
update
set target.[pkAddressID] = source.[pkAddressID]
;
--- Execute SQL Task 3 --- Populate Employee table
Merge [dbo].[Employee] as target
using
(
select [pkAddressID], [Name] from [dbo].[1Stage]
) as source
on source.[pkAddressID] = target.[fkAddressID]
and source.[Name] = target.[Name]
when not matched then
insert ([fkAddressID], [Name])
values (source.[pkAddressID], source.[Name])
;
The short answer is that, out of the box, SSIS isn't built for that. Once you land data in a table, the destination components don't allow for an output stream to be sent.
You could fake this behaviour out by using an OLE DB Command
but your performance will be less than good since it will issue a singleton insert statement for every row that flows through the data flow. Generally, the engine attempts to batch N units of work up and perform bulk, set-based operations on the data to get more throughput. You could also use a Script Component
to perform this. The same performance caveat would still apply.
A better option might be to land your data into a staging table and then use an Execute SQL Task
after your Data flow
and use the OUTPUT clause of the INSERT operation to capture those identities and then patch them into your other table.
The other way to overcome The INSERT statement conflicted with the FOREIGN KEY constraint ...
error while inserting into two related tables with Multicast is to clear option Check constraints
for dependent OLE DB Destination
:
Another approach I would try in with task like yours is to artificially generate the ID field for the parent table. The idea here is knowing the ID ahead so you can assign the foreign key values.
Then instead of using multicast, load the data sequentially: parent, and then child. For the parent table, tick the Keep Identity property (OLEDB Destination).