Tracing and Profiling Java applications in production

*PLEASE NOTE: In order to address quite a bit of hesitation from members wanting to provide first and last name on this site, we are requesting that folks please sign up via Eventbrite. To confirm your seat to this meetup, please RSVP HERE (https://www.eventbrite.com/event/9162195373)

*Please take a moment to review the Check-In Process (https://www.meetup.com/sfjava/pages/Twitter_HQ_-_Meetup_Check-In_Process/) at Twitter to ensure a smooth entry into the event.

http://photos4.meetupstatic.com/photos/event/3/6/d/0/600_289814032.jpeg

For our November meetup, we are extremely happy that Kaushik Srenevasan from Twitter offered to tell us how tracing and profiling in applications in production enables Twitter's JVMs to scale to 1000s requests/second.

The ability to understand the behavior of a software system so as to be able to answer arbitrary questions about its health, while important has always been a challenge for software developers. Typical diagnostic tools tend to either be myopic in their scope or impose a very high performance overhead for their use. This talk describes a new, low overhead, full stack tool (based on the Linux perf profiler and infrastructure built into the Hotspot JVM) we've built at Twitter to solve this problem of dynamically profiling and tracing the behavior of applications (including managed runtimes) in production.

The venue, food, and drinks will be provided by Twitter.

About Kaushik Srenevasan

Kaushik is originally from southern India, but spent the last few years in Redmond, WA at Microsoft, building VMs and backend compilers. Efficient implementation of dynamic languages is something he's extremely passionate about and has spent most of his carrier doing - on two different implementations of JavaScript and the .NET Common Language Runtime. He's currently the Lead on the Diagnostics Team at Twitter where he hacks on the Hotspot JVM, among other things, improving its observability.