Types of Big Data Implementation

¿How can you make a Big Data Implementation? There is no single answer to this question because it will depend on:

– The business intelligence maturity of the company.

– The infrastructure that the organization has for deployment.

– Your level of knowledge of the new environment, your experience (Data science).

– Your approach, more innovative or more traditional, which will unequivocally determine your choice.

In any case, we can say that, in general, the companies, according to its configuration and depending on the aforementioned factors, are grouped according to three different Big Data implementation models:

– revolutionary

– evolutionary

– Hybrid

The revolutionary implementation of Big Data

Who chooses this how to implement Big Data They have decided to break with everything and start from scratch. It is obvious that all companies are already working, have generated data and have been saving and processing it in their own way. Until first contact with Big Data have made data-driven decisions, but always with a high component of uncertainty and a proportion of facts that today may seem dangerously ridiculous.

This implementation puts them on the path to the best decisions because it gives them the ability to do so., not only in terms of volume but also in terms of analytical power. Although it sounds like magic, take this giant step, qualitatively speaking, requires nothing more and nothing less than moving all available data to the new environment. Y, along with them, reports, the modeling and integration with business processes. The past is behind us and from the first moment everything will start from scratch on this new platform.

There is 3 three technologies that allow it to be carried out: hadoop, the parallel databases and the in memory databases, to real-time analysis. Through their tactics of “divide and conquer”, allow you to work with large volumes of information in a short time and at low cost, complemented with the possibility of transferring everything to the cloud.

* Advantages of this type of implementation:

– Agility in information processing– Since each serving is processed much faster and what you used to take 6 hours now takes minutes.

– Optimization of resources that implies a remarkable cost reduction compared to other methods.

– Flexibility of use thanks to the system of nodes: which also allows you to calculate what you are going to spend and pay only for what is used.

* Disadvantages:

– High costs in terms of skills to which would be added the difficulty of finding candidates with this profile in the market, professionals who know and know how to properly handle this technology, making the most of it.

– Perhaps it can also mean a longer adaptation time until commissioning, if compared to other systems, as it is a totally new method.

Big Data Implementation: the evolutionary method

The evolutionary method is another way to access this world that companies usually choose, perhaps not so pioneering, but they already had a WITH A quite mature. We are talking about organizations that have their Deposit, its visualization and reporting tool and who have been analyzing data for years, old-fashioned but mature.

Settle for it evolutionary method of Big Data implementation It is assumed that, maintaining the current structure, data that had no place in the system is simply aggregated to preprocess from the platform Big data. Thus, the current system can see them and thus they are achieved analysis, that remain of the same or similar type, But now incorporate more data.

The advantage of this option is that your entry threshold has a lower cost, as companies can still continue to use their tool, although the Data extraction and its structuring is in charge Big data. Specific, Big Data becomes an entry to the BI Platform existing. Data is accumulated and analyzed, and the results are sent to the data warehouse.

* Advantages of this system:

– Performance. Models are now fed by increasingly diversified data.

– Volume. see multiplies the volume of information you get, thanks to Big data.

– Savings. Both in time of implantation, as in costs, as WITH A existing.

* Disadvantages:

– The speed from one end to the other will always be limited by current BI environment (which is much slower).

– The level of perception is not so broad, since granularity would fail due to WITH A existing.

– This solution is not definitive, because there will come a time when the WITH A does not have the capacity to meet the informational requirements of the organization and has to choose to carry out the complete transfer of the data outside the company.

The hybrid alternative to Big Data deployment

This system alternates the use of one and another technology depending on the objective pursued. For a certain type of information, analysis or user, the WITH A existing, while for much more refined analysis, Like those of predictive type, simulations, etc. would be used Big data. To make it, it would be enough to establish a couple of integration points, obtaining the data of the data warehouse are stored in the scanning engine, which would be seen by the analyzer as a DataMart.

In this option Big data it is not just an annex to the WITH A traditional that now allows you to see data that you could not before, but it is also a platform that serves to make advanced analysis mixing data from the traditional ones that still exist, with the new ones like Social media o la Unstructured data that the BI of before did not contemplate when working.

* Advantage:

– Precision: systems are now fed with structured information, when before they could only access unstructured data.

– Big Analytics. Or what is the same, possibility of making a predictive analysis, creating a statistical model with all the data and identifying causal relationships and correlations; also relying on advanced visualization tools.

– Internet of things: that allows to know in real time everything that is happening anywhere and in relation to any issue.

– Cost effectiveness. Definitely, this option allows you to offer better services than, of course, can also be charged.

* Disadvantages:

– Endurance: of the least evolved companies, technologically speaking, to the implementation of such a model.

– Difficulty: of those who are accustomed to limiting themselves to a type of simply descriptive analysis, when it comes to launching into new possibilities.