Uploaded July 2026 | Updated September 2026, 2 weeks ago
Think Java is too heavy for AWS Lambda? Single-digit adoption rates say yes, but AWS Hero Vadym Kazulkin reveals how modern optimization tools completely eliminate the 3-second cold start penalty.
In this InfoQ video, Vadym Kazulkin breaks down the architectural realities of running enterprise Java in a Serverless ecosystem. He provides an unvarnished, data-driven comparison between AWS SnapStart (with class priming) and GraalVM Native Images. From memory footprints to the internal mechanics of AWS Firecracker microVM caching, this session covers exactly how to achieve sub-second cold starts. Finally, Vadym addresses the future of the ecosystem, including the impact of GraalVM's decoupling from the Java train and the promise of OpenJDK's Project Leyden.
⏱️ Video Timestamps (For Navigation)
00:00 — The Reality of Java Adoption on AWS Lambda
01:45 — Inside a 200-Lambda Production Architecture
02:50 — Anatomy of a Cold Start: Firecracker & Class Loading
05:15 — AWS SnapStart Explained: Managed MicroVM Snapshots
07:30 — Advanced Priming: Eliminating Lazy Loading with CRaC Hooks
11:10 — Performance Benchmark Results (SnapStart vs. Baseline)
13:15 — Tweaking Memory, CPU, and Tiered Compilation
15:20 — How AWS Low-Latency Tiered Caches Work Under the Hood
18:10 — GraalVM Native Image vs. SnapStart: The Trade-offs
21:40 — The Future of Java Serverless: Project Leyden vs. GraalVM
🔗 Transcript available on InfoQ: bit.ly/4ydNalZ
#ServerlessJava #AWSLambda #SoftwareArchitecture
Think Java is too heavy for AWS Lambda? Single-digit adoption rates say yes, but AWS Hero Vadym Kazulkin reveals how modern optimization tools completely eliminate the 3-second cold start penalty.
In this InfoQ video, Vadym Kazulkin breaks down the architectural realities of running enterprise Java in a Serverless ecosystem. He provides an unvarnished, data-driven comparison between AWS SnapStart (with class priming) and GraalVM Native Images. From memory footprints to the internal mechanics of AWS Firecracker microVM caching, this session covers exactly how to achieve sub-second cold starts. Finally, Vadym addresses the future of the ecosystem, including the impact of GraalVM's decoupling from the Java train and the promise of OpenJDK's Project Leyden.
⏱️ Video Timestamps (For Navigation)
00:00 — The Reality of Java Adoption on AWS Lambda
01:45 — Inside a 200-Lambda Production Architecture
02:50 — Anatomy of a Cold Start: Firecracker & Class Loading
05:15 — AWS SnapStart Explained: Managed MicroVM Snapshots
07:30 — Advanced Priming: Eliminating Lazy Loading with CRaC Hooks
11:10 — Performance Benchmark Results (SnapStart vs. Baseline)
13:15 — Tweaking Memory, CPU, and Tiered Compilation
15:20 — How AWS Low-Latency Tiered Caches Work Under the Hood
18:10 — GraalVM Native Image vs. SnapStart: The Trade-offs
21:40 — The Future of Java Serverless: Project Leyden vs. GraalVM
🔗 Transcript available on InfoQ: bit.ly/4ydNalZ
#ServerlessJava #AWSLambda #SoftwareArchitecture
