Database Reference
In-Depth Information
Can we really wait for the data to be loaded into the warehouse only for it
to be sucked out by Analysis Services during cube processing? I am certainly
seeing fewer people wanting to pay that latency price. Once the data has
been integrated, they want their analytical tools to query the data in situ.
Therefore, people are looking at relational online analytical processing
(ROLAP) engines with renewed interest. The Tabular model, offered by SQL
Server Analysis Services, also offers DirectQuery as an alternative.
New Data
It should be no surprise to learn that social media has revolutionized the
kind of insights that users want from their warehouse. Brands can be made
or destroyed by their perceived online presence. A good, responsive
customer support initiative can have a massive multiplier effect to a
company's image. Therefore, sentiment analysis applications are big
business.
What is possibly not so widely recognized is that all these social media data
sources are not “owned” by the enterprise. Companies have been forced
to come out of their shells and relinquish some element of control to a
third-party site. Want to be on Facebook? Then accept that your page is
www.facebook.com/BigBangDataCo
. That's okay for me; my company is
ahem… boutique. However, did you really think you'd ever see Coca-Cola
or Nike doing the same thing? What's more, for the first time a company's
Facebook page or Twitter handle could actually be more a valuable property
and a richer source of analysis than a company's own website.
To complicate matters, the majority of these new properties (Facebook,
Twitter, LinkedIn, and so on) typically exist online. We need to integrate
with them and possibly third-party engines that provide specific
value-addedservices(suchassentiment analysis)toextractmaximum value
and insight from the data.
Social media data is just one example. Let's think about machine data for
a moment. This dataset is huge. There's a myriad of sensors streaming vast
quantities of information, providing a wealth of opportunity for machine
learning and enhanced operations. It doesn't all have to be austere factory
or plant information either. Think about smart metering in your home,
healthcare heart monitoring, and car insurance risk assessment devices
(especially for young drivers), all the way to fitness devices such as Fitbit
or Nike Fuelband that are so popular these days. All are brilliant new data