Subject: | |
From: | |
Reply To: | |
Date: | Tue, 29 Sep 1998 16:18:26 -0700 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Larry Boyd, Wirt, Paul Christidis,
have confered virtually on Data Warehouses...
[all comments snipped to save the archives...]
I would like to make the following comments:
0. Times are changing. Disk is relatively free.
1. The idea of a DW doesn't have to be limited to a vendor
provided solution that generates revenue for the vendor(s).
Don't dismiss the concept just because some vendors have
built a business around the idea.
2. Some organizations have little time opportunity to add indexes,
automatic masters, or other changes to their production OLTP
databases. Having a read-only copy that can be taken offline
periodically is a great benefit.
3. Not all reporting is done using DB performance friendly software.
There is a potential performance risk giving pc client s/w to an
end-user and allowing them to use sql using odbc against a
production DB. This can be alleviated by pointing them to a
reporting DB. If they mess up they don't affect the production DB.
4. A lot of DBs are provided by 3'rd party software vendors. You don't
always have the luxury of adjusting the DB for your reporting without
violating your maintenance/support agreements. In this case you could
be best served by developing a DW solution.
5. Even a not so normalized DB is sometimes hard to navigate by the
end-user using a generate reporting tool.
There is more, but you get the idea that I favor and see a need for
DW solutions in various situations. And I don't think you should have
to buy them since you can leverage your existing skills to make them.
And definitely don't buy them because your friend down the block has
one!
But the decision to buy or build should be considered seriously with
all else the organization is trying to accomplish.
Duane Percox ([log in to unmask] v/650.372.0200x608 f/650.372.3386)
http://www.qss.com/ http://qwebs.qss.com/qwebs
http://qwebs.qss.com/faq3k
|
|
|