SSIS and Update-Insert(Upsert) based on a SQL Server Result Set

Question

So on a weekly basis, I have to sweep a sub-set of our Member data. So my query is built to produce this result set. I then have to take that result set and compare it to a data table which contains all that same information that I have provided to a 3rd
party vendor. So I need to take into account the following scenarios...
Obviously, if the Member is new, then I have to Insert its row to what we'll call table 3rdPartyMember and create a row out on a 3rdPartyMemberAuditTrail Table
Individually, I also have to manage if any of the following data fields have changed: Name, Birth Date, Addressing Information, Member Plan, Member Termination...and by Individually what I mean is that if the Last Name has changed, then update its row in
3rdPartyMember and Insert a row to 3rdPartyMemberAuditTrail indicating "Last Name Change"...and similarly if Address Line 1 has changed..."Address Line 1 Change"
So I guess my question is this. Should this be done in one whole Stored Procedure invoked by my SSIS Package or should it be done in pieces?
Create the Temporary Table to house our weekly Member subset extract and result set
Determine if the row exists on 3rdPartyMember Table and if it does not, Insert it
If the row exists, determine if there is a First Name Change...Last Name Change...BirthDate change...Address Line 1 change...etc... If there is a change, Update 3rdPartyMember Table and also Insert a row to 3rdPartyMemberAuditTrail indicating the change
Based on all this, I'm just not sure the right way to approach this...meaning should we create one big Stored Procedure to handle everything or each individual data point in my SSIS as an UpSert based on its lookup. I am new to the SSIS World and don't really
know what the generally accepted practice is per se of creating and running an SSIS Package with multiple data point update steps or doing so in one big Stored Procedure.
Also....if anyone know of some good YouTubes or web sites that would instruct me as to how to go about doing this, I'd GREATLY appreciate it.
Thanks for your review and am hopeful for a reply.

Answer 1

In my opinion using SSIS is quite a pain for Upsert-functionality, especially with your requirement of maintaining an audit trail.  With Google you can probably find 20 different variants all of which are quite complex.  I would go for the procedure
version just to keep my sanity.  It is so easy to test and verify.
I have been using an excellent Uppsert add on from Pragmatic Works Task Factory, but that would not help with the audit trail, though...
http://pragmaticworks.com/Products/Features?Feature=UpsertDestination(BatchUpdateOrInsert)
I recently discovered the CHECKSUM-function in Transact-SQL, it might simplify your code and certainly improve performance, if you have lot's of rows.
Here is an example:
create table member
(id int primary key,
name char(10) not null,
address char(20) not null,
checks int not null);
create table member3p
(id int primary key,
name char(10) not null,
address char(20) not null,
checks int not null);
create index xmember3p on member3p(id,checks);
--create member3p_audit (...);
insert into member values(1,'tom','new york',0);
insert into member values(2,'mary','dallas',0);
insert into member values(3,'marvin','durham',0);
update member
set checks = checksum(name,address);
insert into member3p values(1,'tom','new york',0);
insert into member3p values(2,'mary','chicago',0);
update member3p
set checks = checksum(name,address);
insert into member3p
select *
from member m
where not exists
( select *
from member3p m3p
where m3p.id = m.id );
-- insert into audit...
select m.*,
case
when m.name <> m3p1.name then 'x'
else ' '
end as name_change,
case
when m.address <> m3p1.address then 'x'
else ' '
end as adr_change
from member m,
member3p m3p1
where m.id = m3p1.id and
exists
( select *
from member3p m3p2
where m3p2.id = m.id and
m3p2.checks <> m.checks );
-- loop and update member3p and member_audit-- It's OK to update columns that were not changedupdate member3p set name = @new_name, address = @new_address;