Performance Engineering Meetup #6
Dive into real-world insights, best practices, and hands-on strategies for optimizing software performance!
🚀 Event Overview
Join us for the 6th edition of the Performance Engineering Meetup, focused on real production challenges in modern backend and JVM‑based systems.
🔍 What’s on the agenda
- 👨🏫 Hidden performance bottlenecks
Learn how thread pool queue wait time quietly slows down systems under load — where it appears, how to spot it, and how to fix it in production. - ⚡ Faster Kubernetes scaling for JVMs
Discover how JVM cold starts and slow pod readiness impact traffic spikes, and how CRaC enables near‑instant startup by restoring a warmed‑up JVM. - 📊 Production‑proven insights
Both talks are based on real production experience, concrete lessons learned, and measurable results at scale.
📅 Event Agenda
17:30 – 17:50 ☕ Registration & Networking (welcome coffee)
17:55 – 18:00 🎤 Welcome & Event Kickoff
18:00 – 18:25 👨🏫 Presentation #1: Mateusz Nalepa (Allegro) — Thread Pool Queue Wait Time: Hidden Bottleneck, Even AI Couldn't Find It
Many different systems behave surprisingly similarly: they rely on threads or thread pools, make HTTP requests in blocking or non‑blocking mode, and parse JSON while executing business logic. But sooner or later they begin to slow down — like driving through a city during rush hour, when every street hides another traffic jam. In this presentation, we'll explore where queue wait time actually appears in our systems, how to spot them and how to fix them. All insights come straight from Production.
18:25 – 18:55 ❓ Q&A
18:55 – 19:15 🤝 Networking Break
19:15 – 19:40 ⚡ Presentation #2: Piotr Mroz (commercetools) — JVM CRaC: Accelerating Kubernetes Scaling with Rapid Pod Readiness
Your JVM service starts slowly. Your Kubernetes pods take 30-60 seconds or more to become ready. And during a sudden traffic spike, that delay is all it takes for the remaining pods to get overloaded - leading to errors, degraded customer experience, or even crashes.
So how do you fix it?
During this meetup, Piotr Mróz - Principal Engineer at commercetools - will share how a platform processing $75B in annualized GMV tackled JVM cold start in Kubernetes at scale, and what they found after evaluating multiple approaches.
We'll explore:
⚡ Why JVM startup time matters - and the real cost of slow pod readiness in high-traffic production environments
🔍 The approaches evaluated - from Class Data Sharing (CDS) to GraalVM ahead-of-time compilation, and why CRaC came out on top
🧊 What CRaC is - Coordinated Restore at Checkpoint, a JVM feature that snapshots a warmed-up application and restores super fast
🏗️ How commercetools implemented it - automated load & resilience testing to evaluate CRaC at scale, JVM Class Loader and JIT compilation monitoring, snapshot vs heap sizing, and more
📊 The real results - startup time improved up to 10-20x in production, dramatically reducing the risk of service degradation during scaling events
19:40– 20:00 ❓ Q&A
20:05 – 21:00 🍕 Networking and goodbye
🎙️ Meet the Speakers
- Mateusz Nalepa (Allegro ) Mateusz is a software engineer with nine years of commercial experience. His work focuses on building efficient, well-designed systems, with a particular emphasis on core software engineering principles that drive performance and scalability. He believes that "MS Paint" and the phrase "it depends" are the absolute best tools for explaining complex programming concepts. His favourite AI prompt - "Hey, explain XYZ to me like I’m a 6-year-old"—helps him quickly find a solid point of reference. At work, he enjoys sharing his knowledge and is always eager to onboard new team members. In his free time, he performs in theater and plays the piano.
- Piotr Mróz (commercetools) Principal Engineer helping teams build fast, scalable, and reliable cloud services. He leads performance initiatives across commercetools, focusing on automating load and resilience testing, optimizing JVM performance, improving Kubernetes scalability, and promoting data-driven engineering. Outside of work, he trains in amateur kickboxing and enjoys bouldering.
Photos may be taken during the event and published on the social networks and websites of SII, Sabre, Pega, IG Group, and Allegro for promotional and marketing purposes. Your participation will be understood as consent to the dissemination of your image.
The email address you provide during registration will be stored and used to communicate with you about future events and related initiatives.
Dive into real-world insights, best practices, and hands-on strategies for optimizing software performance!
🚀 Event Overview
Join us for the 6th edition of the Performance Engineering Meetup, focused on real production challenges in modern backend and JVM‑based systems.
🔍 What’s on the agenda
- 👨🏫 Hidden performance bottlenecks
Learn how thread pool queue wait time quietly slows down systems under load — where it appears, how to spot it, and how to fix it in production. - ⚡ Faster Kubernetes scaling for JVMs
Discover how JVM cold starts and slow pod readiness impact traffic spikes, and how CRaC enables near‑instant startup by restoring a warmed‑up JVM. - 📊 Production‑proven insights
Both talks are based on real production experience, concrete lessons learned, and measurable results at scale.
📅 Event Agenda
17:30 – 17:50 ☕ Registration & Networking (welcome coffee)
17:55 – 18:00 🎤 Welcome & Event Kickoff
18:00 – 18:25 👨🏫 Presentation #1: Mateusz Nalepa (Allegro) — Thread Pool Queue Wait Time: Hidden Bottleneck, Even AI Couldn't Find It
Many different systems behave surprisingly similarly: they rely on threads or thread pools, make HTTP requests in blocking or non‑blocking mode, and parse JSON while executing business logic. But sooner or later they begin to slow down — like driving through a city during rush hour, when every street hides another traffic jam. In this presentation, we'll explore where queue wait time actually appears in our systems, how to spot them and how to fix them. All insights come straight from Production.
18:25 – 18:55 ❓ Q&A
18:55 – 19:15 🤝 Networking Break
19:15 – 19:40 ⚡ Presentation #2: Piotr Mroz (commercetools) — JVM CRaC: Accelerating Kubernetes Scaling with Rapid Pod Readiness
Your JVM service starts slowly. Your Kubernetes pods take 30-60 seconds or more to become ready. And during a sudden traffic spike, that delay is all it takes for the remaining pods to get overloaded - leading to errors, degraded customer experience, or even crashes.
So how do you fix it?
During this meetup, Piotr Mróz - Principal Engineer at commercetools - will share how a platform processing $75B in annualized GMV tackled JVM cold start in Kubernetes at scale, and what they found after evaluating multiple approaches.
We'll explore:
⚡ Why JVM startup time matters - and the real cost of slow pod readiness in high-traffic production environments
🔍 The approaches evaluated - from Class Data Sharing (CDS) to GraalVM ahead-of-time compilation, and why CRaC came out on top
🧊 What CRaC is - Coordinated Restore at Checkpoint, a JVM feature that snapshots a warmed-up application and restores super fast
🏗️ How commercetools implemented it - automated load & resilience testing to evaluate CRaC at scale, JVM Class Loader and JIT compilation monitoring, snapshot vs heap sizing, and more
📊 The real results - startup time improved up to 10-20x in production, dramatically reducing the risk of service degradation during scaling events
19:40– 20:00 ❓ Q&A
20:05 – 21:00 🍕 Networking and goodbye
🎙️ Meet the Speakers
- Mateusz Nalepa (Allegro ) Mateusz is a software engineer with nine years of commercial experience. His work focuses on building efficient, well-designed systems, with a particular emphasis on core software engineering principles that drive performance and scalability. He believes that "MS Paint" and the phrase "it depends" are the absolute best tools for explaining complex programming concepts. His favourite AI prompt - "Hey, explain XYZ to me like I’m a 6-year-old"—helps him quickly find a solid point of reference. At work, he enjoys sharing his knowledge and is always eager to onboard new team members. In his free time, he performs in theater and plays the piano.
- Piotr Mróz (commercetools) Principal Engineer helping teams build fast, scalable, and reliable cloud services. He leads performance initiatives across commercetools, focusing on automating load and resilience testing, optimizing JVM performance, improving Kubernetes scalability, and promoting data-driven engineering. Outside of work, he trains in amateur kickboxing and enjoys bouldering.
Photos may be taken during the event and published on the social networks and websites of SII, Sabre, Pega, IG Group, and Allegro for promotional and marketing purposes. Your participation will be understood as consent to the dissemination of your image.
The email address you provide during registration will be stored and used to communicate with you about future events and related initiatives.
Good to know
Highlights
- 2 hours 30 minutes
- In person
- Doors at 5PM
Location
Lubicz Park building A
Lubicz 23
31-503 Kraków
How do you want to get there?
