Manage high volume
Integrating and exploiting big data has recently become strategically essential to businesses.
To be truly agile, transactional master data repositories need to be supplemented by external data from big data sources, so as to be able to run appropriate, near real-time analysis that will make it possible to best anticipate changes in trends and respond to them more quickly. The drawback is that it is obviously a radical departure from the IT models and technologies with which we were previously familiar.
01Points of attention
What are the key features of big data?
The 3 ‘V’s: volume, variety and velocity
The sheer quantity of big data is far greater than the volumes usually processed, and it rises perpetually.
Almost by definition, big data varies a great deal, and is often in unstructured formats.
Big data is generated over far shorter timeframes, and it needs to be collected and processed in near real-time.
Overcoming the technical barriers to storing big data
This is the first point to bear in mind: big data cannot be exploited using the usual database models. Given the volume of data and its wide variety (including multimedia data, etc.), traditional databases (with DBMS or Database Management Systems) are no longer suitable, and attention must turn to “NO SQL” (for “not only SQL”) storage solutions.
These NO SQL databases overcome the limitations of traditional database management systems (DBMS). They emerged at the end of the 2000s, introduced by behemoths such as Google, Amazon, Yahoo! and Facebook, which found themselves constantly facing performance problems when trying to handle the increasing volumes of data they hold.
To implement these warehouses of less structured data, and to be better placed to absorb rising volumes, the answer is to turn to frameworks such as:
Emerged from work by Google and Yahoo !
Highly suited to the exploitation of digital documents;
for example to digitise and index the entire archives of a newspaper.
Particularly useful when data needs to be stored and exploited in graphical forms on the basis of degrees of separation.
Business exploitation of big data and IoT (Internet of Things) data
Obviously, queries in huge data storage silos are not run in the usual way. You cannot search for images in the Google Images database that look like a particular photo in the same way you would look for a customer’s address.
The Blueway platform provides a considerable advantage with NO SQL connectors that are used to write and run queries in these big data silos, analyse the output and then convey data to the business in real time to deliver the right results to the right person at the right time.
The inclusion of big data has become an issue of knowledge and expertise for those companies that enhance and refine the data they just used to exploit. Car manufacturers who regularly record data on car use and behaviour can correlate this with external sources such as weather or traffic congestion statistics, and so measure the influence of external factors.
However, doing so requires working within the constraints of the 3 ‘V’s of big data, analysing a large Volume of data of great Variety at some Velocity to gain complete knowledge.
The final challenge is that of matching big data with the data in the company’s information system.
This is clearly the big data-related business issue that will help companies better understand the customer journey and their interactions with customers, whether in a shopping centre or on an e-commerce website. By linking this data in real time with data in its CRM system or order history in ERP, it is possible to offer customers information that actually meets their needs at the right time instead of just interrupting them with yet another sales promotion.
De-siloing of the information system...