Lawrence Corr talks about Extended Relational Analysis (ERA)
I would like to tell you about a special one off (at the moment) data
modelling course we are hosting next month on
Extended Relational
Analysis. ERA is great technique for entity relationship modelling and
user driven data requirements gathering which was heavily used by the
consulting firm I worked for in the US. I have persuaded Pat Harrell, a
former colleague of mine and one of ERA’s most experienced practitioners
and evangelists to come over and teach us.
You may ask "Why are you running a course on normalisation and 3NF data
modelling when you believe 100% in dimensional modelling for data
warehouse design?"
Well there are several (5 actually) answers:
Firstly, the course is aimed at all data modellers and business
analysts who must capture any data requirements not just those of us who
design data warehouses and data marts. Even experienced relational
modellers who are looking for more repeatable, business user friendly
techniques that remove some of the chance elements from data modelling
should be interested in ERA. It something of a lost art!
Secondly, an increasing trend I have spotted amongst my clients in
the last 18 months is the development of new production applications and
their matching data marts (additions to the data warehouse) in parallel.
When an application and data mart must go live together as no other
reporting solution is being developed it can be both a blessing and a
curse. For the first time we data warehousing folk can influence the data
capture for analytical benefit but we are also at the mercy of
poor/incomplete/evolving data modelling of as yet non-existent data
sources. Our production data analysis colleagues need help.
Thirdly, while normalised database structures “play no part” in the
published presentation layer of a dimensional data warehouse they may
exist in the ETL data staging area. My friend and associate Joe Caserta
(author with Ralph Kimball of The Data Warehouse ETL Toolkit) makes
extensive use of ERA to help him understand merged data sources and design
ETL staging tables.
Fourthly, ERA is the fastest, most consistent, rigorous,
repeatable, visual and user driven approach to relational data modelling I
have come across. When I worked with Pat Harrell at Linc Systems back in
the 90s I was too busy learning dimensional modelling from Ralph Kimball
and applying it on my very first warehousing project to study ERA in depth
or be able to practice it. However, I am increasingly convinced that its
simple yet powerful techniques and emphasis on JAD (Joint Application
Development) workshops can be applied to BI data requirements gathering
instead of the somewhat random, time consuming and often expensive
interview driven approaches we use today.
Fifthly, for anyone who has studied dimensional modelling (with me
or otherwise) or intends to, I will be taking this course myself and the
last optional topic (run by me) will contrast ERA and dimensional
modelling and provide a roadmap/translation for students who currently
work with dimensional models or will be attending one of
our data
warehousing courses in September.
Hopefully the above might persuade you to attend the ERA course or send a
colleague. In addition, Pat and I are offering the 3 day course at a very
special (cost covering) price of £495.
Best wishes
Lawrence
The ERA Workshop will be held 24-26 July at Leeds Malmaison.
Full course details are available here