Poste Italiane powers digital transformation with MongoDB Atlas

Poste Italiane is a national institution in Italy. Through its postal, banking and insurance services, the group touches the lives of almost all Italians. Its 13,000 post offices are used by 35 million Italians each year; each day, six million users access the group's online services and apps.

It is also determinedly forward-looking. In 2018, Poste Italiane established a team called “Cloud Adoption and Technology Architecture” (CAAT) created in 2018 with the aim of helping the company in the path of technological innovation and digital transformation. CAAT is part of the area reporting to the CIO and, as a speed lane in a bimodal IT approach, was a key player in the strategic definition that led the company to be one of the first public cloud spenders in Italy (with agreements with AWS and Microsoft Azure), fostering the adoption of containerized microservice architectures on Kubernetes and the automation of almost all IT delivery. The goal was to balance stability and agility, optimizing IT resources and driving business growth. Today, Poste Italiane is among the biggest spenders in Italy on cloud services, and a leader in microservice and container architectures. MongoDB Atlas is one of the central technologies of CAAT initiative.

“We’re in the process of transforming Poste Italiane into a data-driven company”, says Gaetano Ruggiero, Chief Data Architect and Head of the Data Mesh Architecture and Factory team at Poste Italiane, “in whose decisions and strategies are based on data analysis rather than intuition. We support the adoption of advanced technologies for collecting, processing and analyzing data, as well as a mindset that encourages the best use of data.”

“Data can help us make more informed decisions to achieve greater efficiency, greater customer satisfaction and greater competitiveness in the market.”

In the financial services sector, historically, mainframes have been favored to carry many of the heavy-data workloads.

As part of CAAT, Poste Italiane wanted to explore how to reduce workloads on its mainframes, particularly for very large volumes on products such as BancoPosta, PostePay or PosteVita.

The aim, says Ruggiero, was to establish a more agile and scalable infrastructure, better able to support new, digital workflows.

Poste Italiane analyzed its workloads, finding that 80% were related to read requests such as card balance, movements, and services associated with the user. Only 20% were write requests, actual financial movements involving transactions on the mainframe.

“It only made sense for the financial movements to be hosted on secure but not scalable, on-premises infrastructures like mainframes,” says Ruggiero. “In short, we wanted to protect the mainframes from workloads where 80% of read load could be handled by a fast layer.”

The CAAT team began looking for the best solution to bring the data out of the mainframe environment, while maintaining the lowest possible latency. Specifically, it wanted payment instructions recorded by the transactional architecture to be immediately available on the other DB platforms.

“We soon identified tools such as Apache Kafka, an event broker and change data capture,” says Ruggiero, “but there remained a critical gap. We needed a database that was able to address queries efficiently but that could also manage heterogeneous information in a scalable and flexible way, allocating the data in the most suitable form for the different digital channels.”

MongoDB meets all these requirements and demonstrates high scalability in reading and writing, says Ruggiero. “The closer we looked at MongoDB the more certain we were it would provide a flexible schema on which to de-normalize the data, inserting different information contents into a single view.”

Poste Italiane recognized MongoDB to be the technological reference on the market, already adopted by many large organizations. Furthermore, given Post Italiane’s role as part of Italy’s national infrastructure, MongoDB supports disaster recovery scenarios that required having a replica of the data in a data center capable of intervening in the event of malfunctions.

With the project being of such critical importance, the design and implementation lasted two years. After the go-live, first with online channels followed immediately by Poste Italiane apps, the CAAT team spent a further 12 months carrying out scrupulous fine tuning. By the end of 2022, the solution was managing more than 7,000 readings per second on an on-premises infrastructure of 30 nodes.

Today, Poste Italiane uses MongoDB Atlas primarily to transform heterogeneous sources of data into coherent and consistent aggregates, which offer higher-level content and information. MongoDB also helps replicate content from the main database to free up bandwidth for core activities, such as financial transactions.

Ruggiero says it has proved to be as scalable and responsive as expected: “Certainly, this fluidity and speed have impacted the broader business. Firstly, more effective digital services have improved the customer experience. Secondly, we can ensure greater continuity and resilience across our core activities. We’re using mainframes only for workloads that are most appropriate.”

A document database such as MongoDB Atlas, he adds, is ideal as it allows Poste Italiane to effectively manage flows coming from heterogeneous domains and create self-consistent and self-contained cross-domain aggregations, easily accessible via APIs.

“A de-normalized JSON document adapts perfectly to the need to manage heterogeneous contents in a single view,” he says, “allowing us to improve the performance of operational APIs.”

MongoDB is now a fundamental part of the software stack that was envisaged.

“I remember having great support from the MongoDB team, even though the technology was easy enough to use even in self-service mode,” Ruggiero says. “We’ve since had the opportunity to meet MongoDB specialists in numerous workshops as we look to make greater use of our data.”