Dimensional Modelling Rules, OK      

Dimensional modelling was

a cutting-edge conception;

A way to model business facts

from every direction.

It’s proved a sound approach to take

for warehouses and for marts,

Enabling data to be pulled

for graphs, reports and charts.

 

You start with what your business needs

and know your data source.

Choose the process to be modelled;

importance first, of course.

Collaborate, in your design

with business SMEs.

Identify facts and dimensions;

don’t worry, it’s a breeze!

 

Facts are numbers, gathered from,

or housed within events,

Amounts and counts of different ilks

(except maybe percents?).

As measures of your business growth,

they show its ups and downs,

Its blood pressure, its beating pulse,

its health - in counts and pounds.

 

It’s during database design,

that strange things can appear;

To norms explained in verse above,

they don’t seem to adhere.

 

The first is ‘factless fact tables’!

a paradox, I inferred,

And – ‘Facts don’t have to be measures’?

which seems, at first, absurd!

 

You must ‘declare the grain’ of facts;

the level of their detail.

If that’s too ceremonial

then put it in an email.

Facts inside the same fact table

must have grain the same,

Else any outputs that you get

 might just be rather lame.

 

Dimensions are our ‘views’ of facts,

like Products, Time, and Places;

The context in which facts emerge,

the lenses in their faces.

Facts have many Dimensions, though

they’ll always want one more,

But facts are useless without them,

they’re what lets you explore.

 

It’s no surprise, Dimensions too,

have ‘types’ that might amuse:

My favourite is ‘Degenerate’,

(it’s just a key to use).

There’s ‘Outrigger’, ‘Step’ and ‘Swappable’

(the latter can be ‘Hot’).

Add ‘Junk’, ‘Role-playing’, ‘Shrunken’ too,

By now I’ve lost the plot!

 

The physical architecture

with which your model’s built

Can be a topic too technical,

I’m starting now to wilt.

Star Schema? Snowflake? OLAP Cubes?

or maybe newer breeds?

It doesn’t matter much to you,

you’ve modelled business needs!

 

Your data model build is done,

DRILL DOWN, ROLL-UP, EVERYONE!

Count and aggregate, summarize,

EVERYBODY, SLICE AND DICE!

Group and filter, sort, create,

JOIN THE PARTY, DON’T BE LATE!

Trends and patterns - visualise,

TEARS OF JOY, DELIGHT IN EYES.

Now get your insights every day,

DIMENSIONAL MODELLING RULES, OK!

 

© Ray Cohen, June 2023

Next
Next

Entities - Now There’s a Thing