Architecting Hyper-Scalable Infrastructure for AI and ML-Driven Fintech with Oracle Globally Distributed Database

May 17, 2024 | 9 minute read
Deeksha Sehgal
Senior Product Manager
Text Size 100%:

Introduction

Digital technologies are reshaping payments, insurance, wealth management, lending, and many more processes. This digitization over the last few years has built the foundation for innovations in financial technologies such as mobile money, peer-to-peer or marketplace lending, insurance technology, and crypto-assets that have emerged around the globe.

One of the key features for most of these financial enterprise applications is “Build to scale”. The capability of an application to scale is something that all businesses desire, as it allows an accelerating number of users to onboard onto the platform. Without scalability, many issues in service disruption, performance, slower response times, and loss of users might surface. These factors impact a firm’s revenues, making scalability critical.

The case study that we discuss shows the need for fintech companies to scale, and how Oracle Globally Distributed Database helps them to fill that technological gap and attain scalable design embedded at their core.

Proliferating scalability in financial firms 

Technology innovation is creating new business opportunities for the finance vertical, and among all other disruptive technologies, hyper-scale computing has been the most transformative innovation of many years. The global datasphere, a term coined by IDC to describe all of the data “out there,” is projected to keep growing. The firm estimates that in 2025, the world will create and replicate 163ZB of data, representing a tenfold increase from the data created in 2016. This hypergrowth is the outcome of an evolution of computing that goes back decades. Another report mentions that 39% of banks believe their digital transformation exercise has fast-tracked their product and service delivery. Financial enterprises need to think ahead and plan for their growth – a roadmap where scalability should be at the top of the agenda.

The COVID-19 pandemic drove the adoption of scalable platforms across industries, including banking and financial services. Post-pandemic, financial institutions invest in hyper-scale data solutions to address infrastructure challenges. Scalable solutions enable fintech to meet increasing demands during crises, expand client reach, diversify offerings, and boost profitability. Fintech enterprises achieving hyper scalability can reshape the future of finance.

 

Navigating the Challenges With Scalability

Need to hyperscale and unleash the real value of data

Ever-growing data is the heart of financial firms, and the sense of urgency to develop a comprehensive scaling strategy is important to them. To harness the true potential of data and facilitate hyperscale growth, fintech companies are increasingly turning to distributed databases coupled with cutting-edge AI and ML capabilities.

These solutions empower fintech with improved operations, encompassing fraud prevention, personalized services, payment gateways, risk assessment, and regulatory compliance, providing benefits to companies and their clients. Oracle Globally Distributed Database facilitates hyperscaling to address the surging demands and data-intensive nature of fintech operations.

The solution offered:

Oracle Globally Distributed Database is a distributed deployment of the Oracle Database presented as a single logical database. It has all the Oracle Database features, plus additional capabilities when used with geographic distribution and replication. It thereby combines the benefits of familiar, relational SQL—consistency and reliability—with those of NoSQL—easy scaling and global reach. Key benefits include:

  • Distributes segments of a data set across many databases (shards) on different computers, on-premises, or in the cloud.

  • Shards can use different cloud providers (multi-cloud strategy) and replicas of a shard can be in another cloud or on-premises.

  • Online re-sharding allows you to move data between clouds, or to and from the cloud and on-premises.

  • Add shards online to increase database size and throughput, delivering online elasticity.

  • Delivers globally distributed, linearly scalable, multi-model databases without specialized hardware or software

  • Enables strong consistency, full power of SQL, support for structured and unstructured data, and the Oracle Database ecosystem

  • Built-in Raft replication to provide a consensus-based, high-performance, low-overhead availability solution, with distributed replicas and fast failover with zero data loss

  • Provides flexibility for hybrid or multi-cloud strategies

  • It provides easy access to Oracle's parallelized, scalable in-database Oracle Advanced Analytics' ML algorithms enabling customers to form predictive insights

 

Empowering a US based fintech achieve hyper scalability in their architecture with Oracle Globally Distributed Database

Client's Overview:

  • Provider offers cutting-edge solutions in fraud prevention, digital identity, device intelligence, chargeback management, and payment gateways. Their Solutions empower businesses to reduce fraud-related expenses, simplify operations, and drive revenue growth

  • Their multi-layered risk platform utilizes machine learning and a community-based reputation database for effective risk management across the customer journey. It covers account setup, authentication, activity monitoring, payments, and dispute handling

  • Customer’s environment involved a database having data of hundreds of terabytes, which had become increasingly difficult to manage efficiently. As their data grew, backup and recovery processes took longer and became cumbersome. The customer realized that their monolithic data infrastructure was no longer sustainable and sought to break free from these limitations and transition to a hyper-scale solution that would meet their evolving data needs.

Technological Overview:

  • Oracle is the foundation for transactional payment databases, spanning multiple instances from Release 11g and onward

  • Existing technologies include Partitioning, Advanced Compression, Oracle Active Data Guard, Oracle GoldenGate, and Advanced Security

  • Data intensive OLTP applications with strict SLAs

  • Data warehouse databases support reporting needs

Customer Requirements:

  • Robust architecture with extreme availability, performance, and scalability

  • Flexible scale-out solution to meet rising customer expectations and workload demands

  • Scalable, reliable, highly available infrastructure with industry-standard security and auditing

  • Required a shift to distributed infrastructure, emphasizing the need to break the monolith and eliminate single points of failure during a catastrophic crash

  • Minimal code changes to support complex joins in the application

  • Real-time query execution within 1 to 2 milliseconds

  • Managing a 200-terabyte database became challenging, leading to extended backup times

  • Efficiently handling 200 terabytes of data is crucial for effective fraud detection

  • Integration of AI/ML capabilities for enhanced fraud detection and real-time analytics to support proactive decision-making

Choosing Oracle Globally Distributed Database:

The customer explored various NoSQL and distributed databases, grappling with challenges related to data sharding, synchronous cross-region replication, and the need for application rewriting. Despite evaluating numerous NoSQL options, they ultimately chose Oracle Globally Distributed Database for its ability to provide scalability akin to NoSQL databases while preserving enterprise-grade features such as data consistency, transactions, joins, atomicity, security, and availability, etc. Oracle Globally Distributed Database offers a unified and seamlessly integrated solution, a departure from the modular approach of other new distributed databases that employ a mix of disparate software modules. With a mature Globally Distributed Database, full SQL support, a mature storage engine, and multiple high-availability options, including Raft replication, Oracle fulfilled all requirements, helping the customer achieve their desired objectives.

Deployment with Oracle Globally Distributed Database:        

  • Implemented User-defined distribution methodology to a group of customer(s) (shard isolation) with a common data model using customer ID as a sharding key

  • Horizontal data partitioning into four shards to maximize performance in OLTP, Batch, and Reporting

  • Strict data sovereignty enforcement to prevent cross-region data leaks

  • Simplified architecture with smaller databases for manageability

  • Deployment of two Oracle Data Guard replicas for each shard: one local with synchronous replication and one remote with asynchronous replication

  • Deployment of a Global Service Manager (GSM) in the primary region, paired with local and remote standby hosts for improved system availability and reliability

  • Utilization of Oracle's AI and ML capabilities on massive data volumes for real-time credit card fraud detection with up-to-date data access for real-time decision-making

 

Architecture

                                                                           Architecture

   

Maximizing Efficiency with Oracle Globally Distributed Database:

Our solutions efficiently addressed customer pain points with Oracle Globally Distributed Database, enhancing core components for excellence.

Linear Scalability:

  • Enabled linear scaling of Oracle Databases along with the applications 

  • Scalability of data for the downstream system by assigning each shard to stream data, hence multiplexing the jobs

  • Scaling up and out of OLTP databases, and customers turned out to use a lower number of shards based on workload demand

Better Fault Isolation:

  • Only a part of the platform is affected during an accidental database crash. No impact on other areas

  • Brings strong multi-version concurrency control, data protection, and security for the processing of high volume of complex transactions in customer applications

Value Additions:

  • Utilized essential Oracle features including data consistency, security, availability, performance optimizer, centralized backup and recovery inherent in Oracle Globally Distributed Database

Unlocking Business Benefits: Modernizing the Technology Stack with NoSQL like Benefits and Breaking the Monolith

  • Minimized the impact of catastrophic crashes

  • Enabled resiliency orchestration to help enterprises by providing better fault isolation, as planned/unplanned downtime within one region does not impact other customers

  • Better application SLAs

  • Ability to isolate and manage heavy I/O customers

  • More control over the data placement 

  • Improved customer services

  • Leveraging Oracle's AI and ML capabilities, the customer achieved rapid transaction validation (within mere milliseconds), ensuring a seamless customer experience and trust-building

Summary:

At its core, Oracle is the trusted database partner for financial institutions, enabling them to extract the utmost value from their systems. This invaluable partnership empowers enterprises to embrace next-generation, scalable solutions optimizing their existing investments and  maximizing performance. By adopting Oracle Globally Distributed Database, fintech companies can effectively address their scalability requirements and gain the competitive edge needed to thrive in the rapidly evolving financial technology landscape.

 

Addressing the migration, performance, availability, and regulatory challenges with Oracle Globally Distributed Database

What do you need?

How you can achieve it with Oracle Globally Distributed Database

Hybrid Cloud

Explore Explore Terraform-based Oracle Globally Distributed Database deployment : On-premises and Oracle Cloud Infrastructure

Multi-cloud

Read Deploying linearly scalable Oracle sharded databases across multi-cloud (Oracle Cloud, Microsoft Azure, and Amazon Web Services)

Seamless Data migration

Read Migrating to a Sharded Database

Linear Scalability

Read How Oracle BlueKai Data Management Platform scales to 1 Million transactions per second with Oracle Globally Distributed Database deployed in Oracle Cloud Infrastructure

Data Sovereignty & Data Proximity

Read How to Achieve Data Sovereignty with Oracle Globally Distributed Database

Availability

Read Replication in Oracle Globally Distributed Database

 

 

Deeksha Sehgal

Senior Product Manager

Deeksha Sehgal is a Senior Product Manager in Oracle with over 12 years of experience in IT Infrastructure services across product management, GTM strategy, technical design and integration, service delivery, database consulting, and hybrid/multi-cloud solutions. In her current role, she develops and effectively positions Oracle's Globally Distributed Database services as a standout offering among other products in the market. Her proficiency in market research and product analysis plays a key role in propelling the product's success.


Previous Post

GenAI RAG Likes Explicit Relationships: Use Graphs!

Phil Cannata | 11 min read

Next Post


Don’t Be Fooled by Misleading Data Egress Announcements

Steve Kilgore | 10 min read