Stay organized with collections Save and categorize content based on your preferences.

Boa Vista chooses Google Cloud to migrate all its information to the cloud

About Boa Vista Serviços

Boa Vista Serviços is a company combining analytical intelligence and cutting-edge technology to transform data into solutions to the challenges faced by its customers and consumers.

Industries: Financial Services & Insurance
Location: Brazil

Tell us your challenge. We're here to help.

Contact us

The initiative is a leap forward in the company’s digital transformation, leveraging data analytics, and product creation and refinement.

Results

  • Ninefold increase in the number of analytical models created
  • Time to ingest large volumes of data reduced from up to 24+ hours to a few minutes
  • Ability to train models between three and five times longer than with the previous environment
  • Automated 1,000+ hours of operational work from the infrastructure team
  • Price cuts for nearly every product—Bluebox now costs around 15% of its original value

Model processing up to 20x faster

Data drives the work of Boa Vista Serviços’s 800+ employees. As one of the main analytical-intelligence and credit-bureau players in Brazil, the company has an extensive database of files on approximately 280 million people and corporations. Using this invaluable information, the company develops solutions to support credit activities and protection for its nearly 21,000 active customers, including financial institutions, fintechs, and Brazil’s largest corporate and retail groups.

Such a robust business needs cutting-edge technology to meet market needs quickly, deliver increasingly relevant products to its customers, and keep a competitive edge. However, the company’s previous technology infrastructure was an obstacle to achieving those goals. Physical servers located in three data centers were starting to prevent the company from attaining the speed and ability to innovate it needed.

As the team had limited scalability for its resources and processes, they could not train many analytical models at the same time or implement modern machine-learning techniques. In fact, the sales of one of its products were about to grind to a halt because the company’s processing capacity had reached its peak. This had a direct impact on how the business was able to expand.

Consequently, the company decided to embrace digital transformation by migrating all its information from physical servers to the cloud. The teams’ organizational model also switched from a traditional approach—with development, architecture, product, and other areas working separately—to a squad-based model. Boa Vista’s 20-plus squads are formed by a cross-section of professionals to focus more efficiently on several technology and business aspects and challenges.

To put that initiative into practice, Boa Vista needed a cloud provider whose technology and solutions were the best match for its business, especially its analytical intelligence requirements. The company chose Google Cloud as a partner and the destination of all its applications.

Building a data lake in Google Cloud

The first area to be migrated was big data, where there was an urgent need to increase processing capacity. The data processing tool previously used with the data center had resource-sizing and governance constraints leading some employees to duplicate the database and use it independently to develop their products.

This generated multiple versions of the databases with different languages and technologies and increased storage costs for needlessly repeated data. The solution was to create a centralized data lake in the cloud to deliver more robust processing, high scalability, and improved information governance.

With support from Google Cloud’s team, Boa Vista was able to adapt its big data infrastructure to the cloud in a modern, efficient way. The company decided not to migrate the environment as it was in the previous tool, but rebuild the processes using Google Cloud’s solutions. It also created a cloud-focused team, which helped speed up the process.

In just four months, the previous tool’s entire original environment was already running natively in Google Cloud. Most of the company’s most strategic databases have already been migrated, and 80% of all databases are projected to be in the data lake by late 2021. Every new product is also now launched integrated in the platform.

100% cloud-native architecture

The main tools in Boa Vista’s data lake include: Cloud Storage, used as the foundation for the data mesh; Dataproc, which runs the mass information ingestion processes; and Vertex AI, used as the access platform for data scientists.

But the highlight is BigQuery, where the structured databases are stored. Besides enabling a quantum leap forward in information governance, this solution shines a light on consumption costs for every product and offers security in data availability, as products only receive the fields relevant for their operation.

Another crucial security aspect for Boa Vista is process centralization so that data-masking techniques can be applied and synthetic databases created to strengthen sensitive information protection. Google Cloud’s VPC Service Control is used to give projects an exfiltration layer. Not even a project’s managers can move the data, as permissioning is restricted to the organization level.

“Creating this environment helped us become one of the first companies in Brazil to achieve ISO 27701 certification on data privacy.”

Luiz Coutinho, Head of Architecture and Security, Boa Vista

The cloud-native architecture reduced the time to ingest large volumes of data from over 24 hours to a few minutes. The company also no longer needs to replicate data in multiple databases to support different products. It is now available centralized in the data lake, with no performance loss.

Hundreds of VMs migrated to the cloud

Another resource category migrated to the cloud was virtual machines (VMs). Using the Migrate to Virtual Machines solution, the team moved nearly 500 VMs encompassing about 7,500 processors to Compute Engine in four months. For the products that needed to be rebuilt, the company chose not to migrate them as VMs, but directly to Google Cloud’s data-analytics tools such as BigQuery and Dataproc.

“We had VMs that hindered our ability to implement or create models or forced us to make them more slowly than we do now. By taking them to Google Cloud, either through VMs or products, we achieved very high scalability,” says Ricardo Orlando, Boa Vista’s CTO.

Every base and component in the applications in the company’s VMs were created through automation, with modules integrating with Google Cloud’s technologies. This strategy delivered more replicability, making it easier to migrate existing products and create new ones. More than 1,000 hours of operational work from Boa Vista’s infrastructure team were automated thanks to the new environment, which also made it possible to generate over 10,000 artifacts in Google Cloud using automation in 2020.

The process was closely monitored by Google Cloud’s team of experts. “Google Cloud has been working very closely with our engineers to deliver the best solution in the cloud,” says Lucas Guedes, the company’s Business VP. “This ensures we can meet all our daily demands and solve any issues we may find to achieve the speed we need.”

New possibilities for data analytics

Migrating to the cloud has directly impacted Boa Vista’s analytics department. Its team can now parallelize the processing of its analytical models without limits, using complex machine-learning techniques, which has increased the product’s operational efficiency. Model processing is up to 20x faster in some cases.

“We have gained a lot of processing power, which has significantly increased the potential of the people working here. The team’s performance capacity as a whole has greatly increased as a result.”

Ricardo Orlando, CTO, Boa Vista

In addition, the ease of access to the data has led to a ninefold increase in the number of models delivered by the team compared with the previous infrastructure. Fifty new models are being created in 2021 alone - meaning 50 new products can be developed using them. The team can also spend more time on training to make more advanced and precise models, which improve product quality and end-user experience.

“Currently, every model is trained between three and five times longer than before. We spare no effort to ensure the best possible performance. Several clients have told us that our products’ precision has us surging ahead of the competition,” says Luiz Coutinho, Head of Architecture and Security.

The new infrastructure allows provisioning new environments in minutes instead of months and scaling the infrastructure according to demand and available financial resources. The costs of migrated products were also reduced. Boa Vista Bluebox, for instance, was reorchestrated in containers in Google Kubernetes Engine (GKE) and started using BigQuery as its database. Now, it costs 15% of its original value.

More agile go-to-market

Analytika is one of the new products launched after migrating. Its development began in 2020 and it is being tested with a customer. As a high-complexity solution, it replicates Boa Vista’s modeling platform so customers can use it to create their own models and solutions, taking every regulation and security issue into account. “This product would be unthinkable if we were not in the cloud,” says the CTO.

Another important launch was the Vulnerability Index (Índice de Vulnerabilidade), created in March 2020 and made available the following month, to help customers make better decisions during the crisis caused by the COVID-19 pandemic. This was possible only because of the ability to develop models and scores more quickly.

“We managed to create a score and make it available for customers to test it and start using it in a month. In the past, it used to take us three months to do that. That more agile go-to-market also helps our customers access better products that reach the market faster.”

Lucas Guedes, Business VP, Boa Vista

De-identified data catalogs of the products were created in the data lake to make it easier for business professionals to access the information. Besides being more convenient for them, this initiative also gave them more autonomy and boosted input in the technology and business sections. This has enabled their squads to pool efforts to create innovative products that keep on surprising Boa Vista’s customers and meeting all their needs.

Tell us your challenge. We're here to help.

Contact us

About Boa Vista Serviços

Boa Vista Serviços is a company combining analytical intelligence and cutting-edge technology to transform data into solutions to the challenges faced by its customers and consumers.

Industries: Financial Services & Insurance
Location: Brazil