What is Datawarehouse and How can we take benefit out from it. ??

Before explaining a data warehouse you need to think of
the opposite, a transaction system. If your company sells
things then you will likely have a transaction system which
stores details of all your customers, what they have bought,
how much they owe, what products you are selling etc.

But this transaction system is ever changing, and it is almost
impossible to find out if, for example, you are selling more this
month that you did last month, or if you are owed more money
this month than last month, or if the sales of a certain product
has increased because of the marketing campaign you ran in
the north for the last three months.

So with a data warehouse you take a 'snapshot' of the data at
a particular time:- hourly, daily, weekly, monthly etc. Gradually
you build up a history of what was in your transaction database
and you can then perform 'business intelligence' on that data to
try to help you run your company better.

This business intelligence is done with 'query and reporting' tools
of which there are many, but products such as Excel, Access, Brio,
Business Objects, Crystal Reports, Cognos etc can all be used.

You can also aggregate and summarize the data as you go along,
so that users do do not see all the detail, but just a summary of all
the product sold in the north, or the west and so on.

Of course you store this data warehouse on a different machine,
and in a different database server. This is because you do not
want to impact the performance of your transaction system when
people are doing queries on the data warehouse.

Now the hard part is deciding what data to put in your data
warehouse, how often to put it there, even finding out what all
your data means (many companies do not actually know what
all the data in their database means).

You try to find out by asking 'end users' what they want to find
out about, what queries they want to run. Often, of course, they
do not know, so it has to be trial and error. You give them some
data and let them start, then they ask for more or different data,
and so it goes on.

Building a data warehuse can be a long and complex job (just
finding out what data a company has, where it is, and how to
access it, can be a major job).

There are plenty of books about data warehouse if you look
for them

Alan

Hi Alan/All,

do not know, so it has to be trial and error. You give them some
data and let them start, then they ask for more or different data,
and so it goes on.


Just my 2 cents worth.......

We stopped approaching DWs like this around 1993. And 12 years is a
long time in the IT industry....

Experience has shown us that asking the business users what they want
will almost certainly lead to poor results.

Why?
1. they don't konw what they want and tell you. That leads no-where.
2. They don't know what they want and the do not want to tell you so
they tell you they want 'everything and they will sort out what they
need.'

A 'trial and error' approach to building DWs is expensive and
slow.....it can also be frustrating for all involved...

As early as 1993 some folks were working on developing techniques to
determine what it was that was needed by the business. I wrote an
article about some of these techniques years ago. You can read it here.

http://www.intelligententerprise.com...questid=276869


The more complete article is in my newsletters download.

Experience tells us that if we use people on BI projects who span the
divide of understanding the business while being able to deliver
any/all components of the IT side of the project we can deliver
remarkably valuable projects....even if the industry is new to the
consultant...

A while ago I played the Business Consultant role for an air cargo
company. I had never done air cargo before but using these styles of
techniques we were able to interview about 20 business managers in 2
weeks and took another week to prepare the outline of the requirements
for the next 12-24 months.....we then built a prototype and showed it
to the business managers......they were amazed that we seemed to 'know
what it was they needed' without ever directly asking them.

These techniques have evolved over the last 10 years to the point now
where we are able to develop extremely valauble BI Systems with not a
lot of input from the business managers....nowadays, experienced BI
consultants have a whole raft of 'things to do' and the real question
for the business is 'in which order do you want to approach this list'.

Anyone doing a BI project today starting from asking the business what
they want is limiting the value that could possibly be derived from the
project...

LIke I said....just my 2 cents worth.

Best Regards
Peter Nolan
www.peternolan.com