German insurers have recognized Big Data Analytics as a field of action
The evaluations are currently widely supported on host and SAN infrastructures. These
business intelligence infrastructures, however, offer little scalability, and they are limited
in their performance and stability with soaring amounts of data
The use of modern, x86-based components business intelligence architectures provides
insurance the opportunity to make data analysis more effectively and efficiently
The amount of data in the network increases exponentially. The importance of the evaluation of these large amounts of data - Big Data Analytics - plays especially for insurance a major role. In Germany, the insurance companies have already recognized the opportunities of Big Data Analytics: 21% of insurers are already using big data, a further 13% seriously plan to introduce it. Thus, the industry is taking a leading position in Germany (https://de.statista.com/infografik/3973/big-data-im-finanzdienstleistungssektor/). Priority is given to the information about their own company as well as the sales or broker organizations for insurers. The analyzed business data spans the entire value chain: from product development and marketing, through sales control, and finally to claim/benefits processing. Morevover, information gained in inside sales forms the basis for corporate governance and compliance. The evaluations are carried out mainly in mainframe architectures and Storage Area Networks (SAN). The processing is carried out in batch runs via dedicated caches.
Digitalization with the opening of new channels (Omnikanal initiatives, especially in sales) for customer contact caused a massive growth of data volumes. Thus, when using the existing architecture two crucial bottlenecks emerge:
In the extraction of data and its storage on caches, a loading bottleneck occurs, the caches must be multiplied in volume
With the monolithic calculations by the mainframe, a processing bottleneck occurs that still further increases the processing time parallel loading into the cache is usually not possible
Thus, the rapidly growing amounts of data lead to performance and stability issues due to the bottlenecks inherent to this architecture. The cost of the host and SAN operations cannot be influenced much due to the widespread licensing models costs and also often fixed-step costs.
For these reasons, business intelligence architectures consider it now crucial to have scalable, distributed processing of large data. The processing is thereby performed in parallelization and largely on x86 data cores. In this context, both specific hardware such as mainframes as SAN infrastructures can be omitted. This new architecture foundation for Business Intelligence thus provides potential for increasing productivity in the insurance field:
Strengthening the delivery capability of Analytics: rapid deployment with ever increasing data magnitudes and uniformity of the information through detachment trom multiple storages and evaluations. This significantly contributes to peripheral benefits collection of Omnikanal initiatives.
Optimization of costs through the use of standardized industrial components in the data center. The solution of the binding to suppliers of mainframe infrastructure and the associated licensing models allows new degrees of freedom in the cost structure
Improving sustainability with a higher infrastructure scalability
The growing awareness of the business benefits of Big Data is increasingly shared at the levelof senior management. An essential prerequisite for the implementation of the technologies described is the use of a robust access and authorization management, possibly partial anonymization of data and the guarantee of the collection and documentation of the informed consent.
Statista (Publisher): Big Data in the Financial Services Sector, Hamburg 2015
(https://de.statista.com/infografik/3973/big-data-im-finanzdienstleistungssektor/ last 11.15.2015)
Statista (Publisher): Big Data – Statista Dossier, Hamburg 2015
Norbert Gronau (Publisher): Competitive Factor of Analytics, University of Potsdam, Potsdam
2015-2390691.html last 10.07.2015)