AWS Data Lakes and Analytics for Financial Services

What is a data lake?

A data lake is a centralized repository that allows you to store all structured and unstructured data at scale and run flexible analytics such as dashboards, visualizations, big data processing, real-time analytics, and machine learning, to guide better decisions

Source: Amazon Web Services.

Data lake on AWS

Amazon Web Services (AWS) offers several services that can be used to build a data lake. They include:

• Amazon S3: A highly scalable object storage service that can be used to store all your data.
• Amazon EMR: A managed Hadoop and Spark service that can be used to process data in a data lake.
• Amazon Athena: A serverless query service that can be used to analyse data in a data lake.
• Amazon Redshift Spectrum: A fully managed, petabyte-scale data warehouse that can be used to analyse data in a data lake.

Industry-leading financial institutions

Mastercard acquired NuData Security to improve its fraud prevention techniques by using passive biometrics to authenticate account holders’ identities. NuData uses an Amazon S3 data lake to store customer data that it collects and analyzes in real time. By using AWS, NuData is able to aggregate, anonymize, and analyze petabytes of customer data to detect anomalous behavior patterns and protect customers from fraud.

Capital One wanted to leverage machine learning capabilities to provide better fraud detection services for its customers. The bank chose to build a data lake on Amazon S3, enabling it to store and analyse large volumes of data. Using Amazon S3 means the bank is better able to detect and prevent fraud in real time. When suspicious activity occurs, Capital One automatically alerts customers and walks them through how to report instances of fraud.

National Australia Bank (NAB) built its Data Hub data lake to power “Discovery Cloud,” a laboratory for the bank’s data scientists. By building its data lake on AWS, NAB is able to provide full data lineage, access the data in real-time via APIs, and analyse the data using a wide range of AWS or third-party services.

Nasdaq needed to provide greater accessibility to data for both internal users and regulators. By building a data lake on AWS, Nasdaq is able to move an average of 30 billion rows into the cloud everyday (with 60 billion on a peak day), while fulfilling security and regulatory requirements and realizing cost efficiencies.

FINRA Case Study

FINRA (Financial Industry Regulatory Authority) leverages AWS (Amazon Web Services) to build and manage its data lake, a central repository for storing and analysing vast amounts of trade data. This data lake enables FINRA’s analysts to efficiently investigate potential fraud, market manipulation, and insider trading.

Key aspects of FINRA’s AWS data lake:

Data Storage:

FINRA utilizes Amazon S3 (Simple Storage Service) for storing raw, unstructured data in its data lake, allowing for scalability and flexibility.

Data Cataloging and Transformation:

Amazon Glue is used for data cataloging, metadata management, and ETL (Extract, Transform, Load) processes, ensuring data quality and consistency.

Data Analysis:

Amazon Athena, a serverless query engine, allows analysts to perform SQL queries directly on the data lake, facilitating efficient data exploration and discovery.

Benefits:

The data lake enables FINRA to analyze years of historical market data quickly, identify potential violations, and support regulatory oversight effectively.

Scalability and Security:

AWS services provide the scalability and security infrastructure necessary to handle the massive volume of data FINRA processes, according to a blog post on Amazon Web Services.

Data Access:

According to a FINRA document, access to specific datasets may require signing a user agreement, and some are immediately accessible. Firm data access is controlled by the firm, and users should contact their firm’s account administrator for access.

Data Lakes and Analytics on AWS

Source: Google Images

Statistics

We get 3x better price-performance delivered by Amazon Redshift compared to other cloud data warehouses.

As compared to open-source Apache Spark, 3.9x better performance is delivered by Amazon EMR.

Trillions of requests are processed per month by OpenSearch Service.

Hundreds of millions of data integration jobs run on AWS Glue every month.

Conclusion

In this blog, we got introduced to Data Lake on AWS. We looked at various services and components of data lake and analytics solutions on AWS. The key aspect of data lake is the enormous benefit it offers over existing solutions. We looked at different success stories along with references to numeric statistics to support the same.

About CloudThat