Big data processing: the Revelis case

Artificial Intelligence solutions

Big Data processing

Big Data processing has become an essential component to fully leverage the informational potential enclosed in the vast datasets available in the digital era we live in. Efficient data management has become crucial for businesses striving to remain competitive and innovative.

In this article, we will explore the design of a high-reliability infrastructure for Big Data processing, focusing on the successful solution adopted by Revelis.

Big Data Processing: technological landscape

Before delving into the details of the Big Data processing solution adopted by Revelis, it is crucial to understand the fundamental technologies used for managing large amounts of data. The key phrase in this context is “Big Data processing,” a term that will be at the center of our analysis.

Revelis’s technological ecosystem is based on a combination of cutting-edge tools. Communication between the various components of the platform occurs through Apache Kafka, a distributed streaming platform ensuring the reliability and scalability needed to handle large volumes of real-time data.

Java programming language forms the core of the system, supported by Spring Boot and reactive programming to create a microservices ecosystem. This choice not only allows for the development of robust and modular applications but also enables horizontal scalability to adapt to the company’s growth needs.

The overall architecture is cloud-oriented, with all components containerized and managed on a Kubernetes cluster. This choice facilitates resource distribution and management while ensuring greater flexibility in scaling applications in response to load variations.

For Big Data storage, Revelis utilizes Elasticsearch, a highly scalable and versatile distributed search engine. This enables quick access and analysis of data, providing a solid foundation for the implementation of artificial intelligence algorithms and advanced analysis.

Regarding streaming data processing, Revelis relies on Apache Flink, an open-source framework designed to handle real-time data streams with high efficiency and low latency. This choice is crucial for applications requiring real-time analysis and immediate decisions based on incoming data.

The processing of AI algorithms is entrusted to Apache Spark, a distributed processing framework offering powerful computing capabilities for machine learning and predictive analysis. The integration of Spark allows Revelis to implement advanced AI solutions, harnessing the potential of accumulated Big Data.

Designing a high-reliability infrastructure for Big Data processing

Now that we have outlined the basic technological context, let’s explore how Revelis has designed a high-reliability infrastructure for Big Data processing. Below, you can see the general architecture diagram implemented by Revelis.

The platform’s features are delivered through a Spring Boot application, enriched with specific security mechanisms, exposing appropriate API REST necessary for the operation of the Presentation Layer and integration with external platforms wishing to utilize these features.

The entire platform can interact with the Kubernetes cluster on which it is installed and create/manage POD or other types of resources.

The main concepts of the platform (also at the base of the IoT Layer) are Gateways and Devices:

  • A Device is a hardware (e.g., an IoT sensor) or software (any other data source) device capable of transmitting or generating data. Data transmitted by a Device is represented using a unified model internal to the platform. This model is used both to transmit data (from devices) on the internal queue (in this case, the data is serialized in Json format) and to store it on Elasticsearch (again, the data is serialized in JSON format);
  • A Gateway is an ad-hoc application (in the platform, there are already various ready-to-use implementations based on the context) capable of connecting to Devices. When a Gateway connects to a Device:
  • Reads its data.
  • Decodes it into the unified model.
  • Transmits it, in JSON format, on the internal queue (Kafka).

The Data Layer consists of an Elasticsearch cluster (where Big Data is stored) and a relational database necessary for the platform’s operation and data enrichment with additional meta-information.

In the Business Layer, in addition to the REST API layer, all components necessary to provide services such as:

  • Alarm and notification management: through a suitable rule engine, it is possible to configure processing flows for real-time monitoring of certain conditions, upon which alarm messages can be generated and sent to specific channels (email, WhatsApp, web notifications).
  • Business Process Management (BPM): through a suitable graphical environment, it is possible to design and execute any business process, integrating information from any source (e.g., IoT devices).
  • Predictive maintenance: in the case of data from machinery, the platform is equipped with algorithms capable of detecting failures preventively.
  • Anomaly analysis: in the case of monitoring IoT devices, the platform has algorithms capable of understanding whether a given sensor is transmitting correct data or not.
  • Customized analysis: thanks to native support for frameworks and libraries such as Apache Spark,Keras, Tensorflow, etc., the platform allows integrating any type of customized analysis.

Advantages of the solution developed by Revelis

The solution developed by Revelis for Big Data processing offers numerous advantages. Let’s examine them in detail.

Reliable Communication with Apache Kafka

The choice to use Apache Kafka for communication between platform components has proven to be crucial in ensuring the consistency and reliability of exchanged information. Thanks to Kafka, Revelis has implemented a distributed messaging system that handles large volumes of data without compromising speed or security.

Java Programming and Scalable Microservices

The Java programming language, supported by Spring Boot and reactive programming, provided Revelis with the flexibility to develop a highly scalable and fault-tolerant microservices ecosystem. This modular architecture allows Revelis to adapt quickly to market changes and deploy new features without disrupting operational flow.

Cloud-Oriented Architecture with Kubernetes

The adoption of a cloud-oriented architecture, with all components containerized and managed on a Kubernetes cluster, has significantly improved Revelis’s operational flexibility. Automatic scalability and simplified resource management have enabled efficient resource utilization, minimizing downtime and optimizing overall performance.

Efficient storage with Elasticsearch

Elasticsearch has proven to be an ideal option for Revelis’s Big Data storage. Its ability to index and retrieve information quickly from large data sets has allowed the company to efficiently access crucial information for analysis and AI algorithm implementation.

Streaming processing with Apache Flink and AI Processing with Apache Spark

The combined use of Apache Flink for streaming data processing and Apache Spark for AI algorithm processing has allowed Revelis to maximize its analytical capabilities. The constant flow of data is efficiently managed by Flink, enabling real-time analysis, while Spark provides the computational resources needed to implement advanced AI algorithms on extensive datasets.


In conclusion, designing a high-reliability infrastructure for Big Data processing is a crucial challenge for companies seeking to derive maximum benefit from their available data. In the case of Revelis, the combination of advanced technologies such as Apache Kafka, Java, Kubernetes, Elasticsearch, Apache Flink, and Apache Spark has created a robust and scalable environment, allowing the company to provide cutting-edge AI solutions and Big Data analytics.

Revelis’s experience demonstrates that investing in a solid technological infrastructure is essential for addressing the challenges of Big Data management and enabling advanced analysis leading to more informed decisions and more effective business strategies. With Big Data processing at the core of operations, Revelis shows that technological innovation is the key to success in the digital era.

Author: Massimiliano Ruffolo