OLAP Versus Big Data

Very interesting discussion OLAP Versus Big Data from  datawarehouse.ittoolbox.com and very interesting answer Peter J

“It doesn’t have to be Big data vs BI or OLAP.
Big data has become the hot trend and most people have no clue what it
really is or how to actually use it.
Great they can get more data and not do anything with it. It’s the whole
grab as much data and it’ll ‘automagically’ just work for you. Tons of
companies market GUI front ends that don’t do anything for a company if the
data is crap. Sell and market whatever you want, if the data stinks, it
just doesn’t matter. And right now big data is more about the technology
than the business.

Yeah yahoo, google, facebook, linkedin, groupon etc can throw out their big
data usage, their massive work on big data, but those companies rely on
unstructured data, streaming data, near real time metrics, etc. They all
are more or less advertising companies. Without advertising, those
companies make no money. Banks and wall street have a need for big data
because they need tons of data, lots of unstructured data, and they need it
yesterday.

A lot of big data is really just unstructured data. How to get the logs,
the web streams, the emails, the social media status updates, maybe image
and video info and find useful ways to help your business or organization.
It can be a few terabytes or a few petabytes. It’s more about structured
or barely structured over tons of data.

OLAP and BI systems are mostly structured, single versions of truth. Do
it the Kimball way, Inmon, or what not doesn’t matter because you take
different systems, different data, normalize it or denormalize it, wrap it
up into star schemas, and run calculations and aggregations on it and then
build cubes or reports based off of that. It’s all the same in regards to
having one large EDW or a ton of silo data marts. it’s mostly structured
data.

So it’s not about doing OLAP or Big Data, it’s figuring a way to integrate
Big Data, mostly unstructured data, with your BI and MDM systems.
Integrate it all together and it becomes useful. Building one big data
system, another BI system, another MDM system, a bunch of OLTP systems, and
so on just means you’re still doing what’s always been done. And that’s
having a ton of different systems that really don’t talk to one another or
relate to one another.

The technology doesn’t matter. It’s about how to use that data to get
some kind of return(more $$$, savings, better data, reports, charts, etc)
for an organization or business. Starting a big data project just because
it’s hot and you need to do it won’t change the facts. If the data still
stinks, it’ll just be another system nobody really uses. At least not use
it well. There is a big reason why, even after all these years, most people
still use Excel spreadsheets to do half their analytics and reporting
needs. More technology isn’t going to help a secretary or mid level
manager who could care less how you get the data.”