Azul Prime Helps Asia's Largest Digital Payments Platform Handle 15% More Transactions with 20% Less Compute — No Code Changes Required
A production pilot across dozens of Java microservices delivered up to 30% latency improvements across P50-to-P99 percentiles, a 20%+ reduction in application instances, and 15% more concurrent transaction capacity — simply by swapping the JVM, without touching a single line of application code or disrupting existing CI/CD pipelines.
5 min read
One of Asia's largest digital payments platforms has achieved a 15% increase in concurrent transaction capacity and a 20%+ reduction in application instances after piloting Azul Prime — the high-performance Java Virtual Machine (JVM) from Azul, the trusted leader in enterprise Java for today's AI and cloud-first world. The results were achieved across production and staging workloads spanning dozens of interconnected Java microservices — with no application code changes, no re-architecting, and no disruption to existing CI/CD pipelines, even during peak transaction periods.
The platform also recorded up to a 30% latency improvement across P50 through P99 percentiles — meaning not just average latency improved, but the tail latency outliers that cause the most damage in payment systems were dramatically reduced. For a platform processing thousands of transactions per second across a chain of interdependent microservices, that latency profile improvement translates directly into fewer SLA breaches, fewer failed transactions, and a materially better experience for end users at peak load.
Why Java Runtime Performance Is a Critical Payments Infrastructure Problem
Digital payment platforms are among the most demanding Java runtime environments in existence. They must process thousands of requests per second continuously, maintain strict latency SLAs across complex microservice chains, and scale reliably through unpredictable demand spikes — flash sales, payroll cycles, festival spending peaks — without overprovisioning infrastructure that sits idle the rest of the time. In a microservice architecture, the Java runtime is not just one component among many: it is the substrate through which every transaction flows. A JVM that pauses for garbage collection at an inopportune moment, or that fails to optimise hot code paths efficiently, creates tail latency spikes that cascade across the entire service chain.
For this particular platform, the transaction flows spanned dozens of interconnected Java microservices, each of which was a potential source of latency variance. Brief spikes in any single service could trigger SLA breaches and cascade into failed transactions upstream and downstream. The combination of high throughput requirements and interconnected latency dependencies made the JVM choice a first-order infrastructure decision — not a supporting technology choice.
What Azul Prime Does — and Why It Requires No Code Changes
Azul Prime is a drop-in replacement for standard OpenJDK — the open-source Java runtime that the vast majority of enterprise Java applications run on. It requires no changes to application code, no re-architecting of microservice boundaries, and no modifications to CI/CD pipelines. Deployment is as simple as substituting the JVM — a change that can be made incrementally across a microservice fleet, service by service, with full rollback capability at any point.
The performance gains come from Azul Prime's core technology: the C4 (Continuously Concurrent Compacting Collector) garbage collector, which eliminates the stop-the-world pauses that are the primary source of tail latency spikes in standard JVM implementations. While OpenJDK's garbage collectors must periodically pause all application threads to reclaim memory — creating the latency spikes that payment platforms most fear — C4 performs all garbage collection work concurrently with application execution, with no pauses regardless of heap size or allocation rate. The result is consistent, predictable latency at any throughput level — exactly the profile that payment microservice chains require.
Higher throughput per CPU core means that fewer instances are needed to sustain the same peak load — which is how a 20%+ reduction in compute infrastructure was achieved while maintaining existing performance SLAs. For a platform operating at the scale of one of Asia's largest payments processors, that reduction in instance count translates directly into a material reduction in cloud or data centre compute costs.
The Broader Azul Evidence Base — Forrester TEI, Sasol, and Fortune 100 Scale
The payments platform results are the latest in a sequence of quantified Azul Prime outcomes released this month. A new Forrester Total Economic Impact™ (TEI) study — commissioned by Azul and conducted independently by Forrester Consulting, examining six organisations that use Azul Prime — found that those organisations achieved a 129% return on investment and a net present value of $5.7 million over three years. The payback period was less than six months. The study attributed the ROI to three primary drivers: infrastructure cost reduction from higher throughput per core (reducing instance counts), operational efficiency gains from eliminating JVM-related performance incidents, and avoided licensing costs from moving off Oracle Java.
A separate case study released the same week showed that Sasol — a global integrated chemicals and energy company headquartered in South Africa — standardised more than 150 business-critical Java applications on Azul Core, achieving a 92% reduction in Java licensing costs while strengthening security and compliance posture. The migration, completed in partnership with Optim-G, eliminated Oracle Java licensing fees entirely across Sasol's Java estate. Together, the payments platform results, the Forrester TEI findings, and the Sasol migration illustrate the two distinct value levers of the Azul platform: performance optimisation (Azul Prime) and licensing cost elimination (Azul Core).
The Java Modernisation Imperative — and Why the JVM Choice Is Now a Strategic Decision
Azul's customer momentum arrives against a broader backdrop of urgency in the enterprise Java ecosystem. Multiple Java Long-Term Support (LTS) versions are approaching end-of-support in a compressed window: Java 17 in 2029, Java 8 in 2030, Java 21 in 2031, and Java 11 in 2032. Organisations that have allowed technical debt to accumulate across long-lived Java applications — running on versions with years of unused code and dependencies tied to expiring support contracts — face a simultaneous modernisation challenge that will require deliberate JVM strategy decisions, not just version upgrades.
Azul powers mission-critical systems for 36% of the Fortune 100, 50% of the Forbes Top 10 World's Most Valuable Brands, and the world's top 10 financial trading companies — a customer concentration in high-stakes, high-throughput environments that reflects the specific value proposition of a JVM designed for exactly those conditions. For the payments, financial services, and high-volume commerce platforms where Java microservice performance is a direct business constraint, the JVM is no longer an infrastructure commodity. It is a competitive technology choice — and the results from Asia's largest digital payments platform make that case in numbers that infrastructure and finance teams can both understand.
Key Takeaways
- One of Asia's largest digital payments platforms achieved 15% more concurrent transaction capacity, a 20%+ reduction in application instances, and up to 30% latency improvements across P50-to-P99 percentiles by piloting Azul Prime across dozens of production Java microservices — with zero code changes, zero re-architecting, and zero CI/CD pipeline disruption
- Azul Prime's C4 garbage collector eliminates stop-the-world GC pauses entirely — running all memory reclamation work concurrently with application execution regardless of heap size or allocation rate — which removes the primary source of tail latency spikes in Java microservice chains and the cascade failures they cause in interconnected payment flows
- A Forrester Total Economic Impact™ study examining six Azul Prime deployments found organisations achieved 129% ROI and $5.7 million net present value over three years, with payback in under six months — driven by reduced instance counts, eliminated JVM performance incidents, and avoided Oracle Java licensing fees
- Sasol standardised 150+ business-critical Java applications on Azul Core this month, achieving a 92% reduction in Java licensing costs by eliminating Oracle fees — illustrating the two complementary Azul value levers: performance optimisation (Prime) and licensing cost elimination (Core)
- With Java LTS end-of-support dates converging between 2029 and 2032 (Java 17, 8, 21, and 11 respectively), enterprise organisations face a compressed Java modernisation window — and Azul's position powering 36% of the Fortune 100 and the world's top 10 financial trading companies makes the JVM increasingly a strategic technology choice, not a commodity infrastructure decision