Gridfore is a project to develop BigData intelligent platform for real-time BI and campaign management, including predictive decision machine and business services.
Gridfore intends to qualitatively reduce Time-To-Market of developing efficient Campaign Management and Business Analytics services, as well as to provide simple tools for elaboration of business service clouds (product factories) with reliable impact and regress assessment.
Gridfore is focused on telco (MVNO including), banking and retail.
Problem
Knowing the customers, their environment and getting full understanding of their way of life.
The modern retail market is set on three main pillars. These are three main capabilities that currently are not completely used. It is the ability to remotely identify the consumer and interact with him making complex deals, forming effective targeted offers to the consumer in real time, using the mechanisms of geo-tracking and analysis of his Internet and service-consuming activities, accumulating customer’s personal behavior, consumer and social patterns in order to create a specular social network of user interaction, so setting up an efficient machine for targeting products and services.
A full combination of all those pillars can be possessed by telcos, who can accumulate customer data on the deepest possible (hardware) level, and provide effective data processing functionality, such as personal promotion of partner products and services as a complex business model for universal selling machine that makes social cluster segmentation, targeting the customer and provides real-time decision services (scoring, anti-fraud, etc.). That concept can be best used as a platform for cloud-based services provided to virtual mobile network operator (MVNO) where the modern banks are about to migrate to.
However, the existing technological stack does not allow the full use of these features. Although their combination will surely lead to a qualitative change in marketing approach in the retail sector — the market is well aware of it and it is looking for a suitable solution such as guessing on Gartner’s squares, making attempts to use legacy vendor platforms and new open stack, which altogether often provide a limited effect, which can disappoint potential investors in the very idea of such a solution.
Solution
Gridfore intends to reduce qualitatively Time-To-Market of developing efficient and reliable Campaign Management and Business Analytics services, as well as to provide simple tools for elaboration of clouds for business services (product factories) with reliable assessment of functional impact reducing regress risks.
Innovation of the approach is to realize the principle possibility of constructing a solution of this class — to make an active data warehouse that can effectively withstand simultaneous OLAP and OLTP highload, providing an access to actively changing BigData in real time with the minimal possible latency.
The solution design also implies metadata management components (providing support for data structures historization and impact assessment tools), tools for improving data quality and unifying the semantics of data sources, and tools for implementing business logic by DevOps engineers. Having its own developed scripting language (groovy-based DSL) alongside with a live code distribution system, allows it to execute the client application code on a cluster (e.g. private scoring services) and to provide machine learning framework in order to train private models that can be enriched by client’s private data and precomputated predictors.
So Gridfore is the Data Management Platform (DMP) with exceptional abilities to additional monetization, such as:
- charging for access to DMP owner (telco) data and private client data that also can be sold from one customer to another;
- charging for the CPU resources for execution of injected private computation requests;
- charging for the user space for storing private trained models, input data and predictors.
The solution is linearly scalable. It provides hot cluster reconfiguration and full-time failover. It can be flexibly integrated with legacy and modern operating platforms.
Trends
The main trend of the technological stack development, for our opinion, is aimed at the consolidation of operational and analytical services in a universal solution that operates the sole layer of reliable data, and provides unified and effective mechanisms of data processing, quality control, and preventing of functional regression.
Impact control avoids the main disadvantage of the microservice architecture, expressed in a disproportionate increase of regression risks, when the graph of interrelated services becomes more complicated (hundreds and thousands of services for a large corporation).
High TTM can be provided by building a reliable and efficient change management process that supports both the parallel development of business services and the continuous refactoring of the solution components.
This very concept is a solution of a new class for the prospective and current business requirements. Obsolete approaches to loosely coupled system integration, using service buses, batch or streaming data exchange lead to low efficiency of the solution for the business, high latency and unreliability of distributed operations, lack of common semantics and a significant complication of the solution entirely. All mentioned above leads to high TCO, uncertainty for the business in the reliability of the solution itself and the possibility of its effective development, which in turn reduces the willingness of the customers to invest in their technical development.
Technology
Gridfore is the project to develop BigData intelligent platform for real-time BI and campaign management, which includes a predictive decision machine and online business services such as scoring, anti-fraud and marketing services.
Gridfore intends to reduce qualitatively TTM of developing efficient and reliable CM and BI services, as well as to provide simple tools for developing the clouds of business services (product factories) with reliable assessment of functional impact, reducing regress risks.
Gridfore as a product is targeted at telco, trade and financial retail, including virtual operators.
The architectural concept is based on convergent Gridfore OLAP/OLTP solution built on in-memory processing technologies (in-memory data grid) and Hadoop stack. The solution is designed as a hybrid system, based on data warehouse with near-zero data latency, combined with high-efficiency compute grid, that encapsulates high-load ELT batch, streaming processing and real-time computing services.
The project Gridfore team has extensive experience in the implementation of BI solutions. Nowadays the team is developing a platform for a global telco. It is currently contracted by a high-tech retail bank and is under negotiation with several potential customers.
The project Gridfore is interested in technological improvements and elaboration which can be achieved through sharing of ideas in the professional and business community.