Vault Core: Peak Performance, Proven the Right Way
A typical bank processes billions of transactions daily, from retail payments to payroll, and poor performance in its core banking system can lead to slowness, outages, and operational risk. If the system struggles to process a bigger-than-expected volume of daily transactions, it can cause backlogs and even complete outages. Potentially halting all banking operations and causing significant financial and reputational damage, such as financial losses and regulatory non-compliance with fines.
Any bank needs confidence that its core will scale to handle spikes in demand with predictable latency and resilience, especially when unexpected. This confidence can be provided by a rigorous x testing framework that delivers transparent, reliable results that reflect the complexity of real-world banking.
The usual approach is often unclear and lacks reliability. You'll see headline numbers like Transactions per Second (TPS) without any details on what they actually mean, how they were calculated, or if they're even reproducible.
Thought Machine's Vault Core is a high-performance banking platform built for modern financial services. Its performance testing proves that a real-time, cloud-based system can meet the demands of speed and reliability and support the bank at any scale.
In this paper, we will detail our performance testing methodology and the results we've achieved with the Vault platform. We'll also discuss our plans to continue pushing the limits and aim to serve the world’s largest banks.
Our high-integrity performance testing methodology
Our approach is built on a foundation of realism and reproducibility. We pride ourselves on being leaders in transparent, meaningful performance validation. To achieve this, we've developed a comprehensive testing framework that showcases Vault Core's capabilities in real-world production environments, built on the following key pillars:
- Realistic financial products
- Real-world production environments and pre-loaded data
- Clear real-world journeys
- End-to-end reproducibility
- Continuous performance gate
Realistic financial products
Many performance tests are artificial and don't reflect real banking. We solve this by employing three distinct financial product portfolios that cover a wide range of features and use cases found globally. This ensures our tests are grounded in authentic banking scenarios, providing you with insights that are directly applicable to your real-world operations.
Real-world production environments and pre-loaded data
How can you be confident in performance results without a true production environment? Our performance tests are conducted in environments that meticulously replicate real-world production setups. We utilise documented infrastructure configurations on leading cloud providers, including AWS, Azure, and GCP. This ensures our results are directly translatable to your potential deployment, giving you a direct preview of Vault Core's performance in your operational landscape. In fact, you could achieve even better performance by leveraging more advanced infrastructure setups.
The test environment is also pre-loaded with a significant, production-scale dataset, including millions of customers, accounts, and a comprehensive history of transactions. This is crucial for mimicking the data density and complexity of a real-world banking system.
Only general availability releases of Vault Core are deployed onto the environments, and we strictly avoid any experimental code or infrastructure configuration.
Clear, real-world journeys
Vault Core’s performance is measured across seven meaningful test journeys designed to cover a full range of banking operations, from single actions like opening an account to the intense, simultaneous demands of peak high-priority transactions and complex end-of-day batch runs. These test journeys are measured in four different test types, providing a holistic view of the platform's efficiency, scalability, and stability under various loads.
This comprehensive approach empowers a clear assessment of the platform's ability to handle transaction volumes and understand its behaviour under peak conditions. Ultimately, this helps ensure the platform consistently meets your business requirements and provides a reliable user experience, preventing issues like slow response times that could impact customer satisfaction and business operations.
.png)
End-to-end reproducibility
Unlike legacy systems that often rely on low-level, hardwired tests which bypass critical API layers and services, our testing framework ensures every test request passes through the same public APIs, internal services, event streams, and database tables as a live, production transaction. This guarantees that test results are not only realistic but also fully reproducible, giving you a clear and honest representation of how many production accounts can be supported.
.png)
Legacy
- Typically low-level tests and hardwired paths to specific use cases
- Blended journeys (transactions and balance inquiry) with API layers stripped away and services written directly to database
- Black box with mixed read and writers, not producing realistic production results
Vault Core
- Realistic and reproducible tests with full Product Library functionality maintained
- Split journeys for full transparency
- Replicating real-world scenarios, clearly identifying how many production accounts can be supported.
A continuous performance gate
To ensure consistent quality and improvement, every new Vault Core release must pass the target set before it is approved for deployment. This rigorous process guarantees a stable and continuously improving performance baseline with each update, giving you peace of mind.
Our performance results demonstrated
Certified results (baseline)
Building upon our rigorous testing methodology, we produce a certification performance report with every release. Among other things, the report shows the performance results of Vault Core within a Tier 1 banking environment, simulating 30 million live accounts and over a billion transactions—a blend of 15 million current accounts with daily usage and 15 million savings accounts with monthly usage.Our report measures six distinct test journeys against four test types, assessing performance both in isolation and in blended scenarios. This approach simulates live operational conditions, providing an authentic depiction of Vault Core’s power and efficiency in real-world banking.
Below are the results for the key journeys. See the appendix for the full report:
- High-priority transactions (Online Postings): Achieving 4,769 TPS with 95% of responses delivered in under 500ms, ensuring rapid and seamless customer interactions.
- Low-priority transactions (Offline Postings): Demonstrating robust processing with 6,768 TPS, where 95% of operations complete in less than 200ms, underscoring our efficiency in batch processing.
- Account opening: Facilitating expansion with 2,000 TPS for account openings, maintaining a 95th percentile response time of under 1,500ms, and empowering swift client onboarding.
- Live balance enquiry: Providing immediate financial insights with 15,035 TPS for live balance enquiries, with 95% of responses delivered in under 200ms, enhancing real-time financial visibility.
- End-of-day processing: Consolidating daily operations in just 90 minutes, processing at a rate of up to 20,676 TPS, ensuring timely and accurate financial reconciliation.
Lab studies (beyond certification)
Beyond standard certification, Thought Machine is committed to continuously pushing the boundaries of Vault Core’s capabilities. In a Tier 1 banking environment with 70 million live accounts and 28 billion transactions, we have successfully developed and implemented the following showcases for two of our Tier 1 clients. This underscores our dedication to innovation and our pursuit of unparalleled performance in the financial technology sector.
Pushing throughput to the limit (TPS)
In this lab study, we demonstrated the maximum throughput achievable with Vault Core for a bank with a very high volume of accounts (70 million). Blended tests showed a sustained capacity of 8,000 TPS for high-priority transactions under 151ms, and 8,000 TPS for live balance enquiries under 42ms (95% percentile).
Supporting Very High Volume Accounts
We define a high-volume account (HVA) as an account with more than 50,000 daily transactions. This lab study demonstrated that Vault Core can sustain performance levels comparable to certified ones in an environment with 70 million accounts, including 32 HVAs with 50,000 daily transactions each and three more with 140,000 daily transactions each. This demonstrates the platform’s ability to support enterprise banks.
Conclusion
Vault Core's meticulously engineered architecture and rigorous testing methodology solidify its position as an industry-leading core banking solution. The detailed test results and real-world showcases demonstrate its capacity to handle the immense scale and complexity of modern banking operations with industry-leading throughput, low latency, and reliability.
By choosing Vault Core, financial institutions gain a platform that not only meets today's performance demands but is also engineered to scale for the future. Global banks such as JPMorgan Chase, Standard Chartered, Intesa Sanpaolo, and Lloyds have already chosen Vault Core as their strategic core banking platform, investing in a long-term partnership with Thought Machine. This partnership will continue to uphold the highest engineering standards and ensure their performance capabilities evolve with their ambitions.
Download the whitepaper below for more data and methodology in the Appendix.
Unsere Pressemappe herunterladen
Hauptsitz:
7 Herbrand Street
London, WC1N 1EX
United Kingdom