• Tales of Instrumentation

    Uber

    After a bit of a break, our Distributed Tracing - NYC meetup is back at the Uber NYC office! Refreshments will be available at 6:20 p.m. and the talks will begin at 7:00 p.m. Anyone involved in distributed tracing knows that the journey starts with getting high quality instrumentation into your applications. So join us for some Tales of Instrumentation. Agenda: 6:30: Networking 7:00: Talk 1 - Instrumentation: breadcrumbs for a dark path in the future 7:45: Talk 2 - Introduction to OpenTelemetry 8:30: Networking 9:00: Lights out ** Instrumentation: breadcrumbs for a dark path in the future ** Presenter: José Carlos Chávez (https://github.com/jcchavezs) In observability, instrumentation is the task of adding signal emitters in our code for further analysis or debugging. Instrumenting a service is the first step towards understanding a system state while operating it but in practice it really depends on how much the libraries one consume can be instrumented. In this talk we will explore the importance of instrumentable code, patterns for writing instrumentable libraries, means for instrumentation, trade offs and lessons learned on the topic. José Carlos Chávez is a Software Engineer at Expedia, a Zipkin team member and a Mathematics student at the University of Barcelona. He enjoys working with APIs and distributed systems, and is the author of official libraries for Zipkin and OpenTracing. When not working with code, you can find him sipping on craft beers. ** Introduction to OpenTelemetry ** Presenter: Austin Parker (https://github.com/austinlparker) You may have heard of OpenTelemetry, the next major release of OpenTracing and OpenCensus, but you may wonder what it actually is, and how it's going to impact you. In this talk, you'll learn about what led to this project, how it builds on existing open source observability frameworks, and how you can join the community and get involved with the next generation of open source observability. Austin Parker is an Open Source Engineer at LightStep and a maintainer on the OpenTelemetry and OpenTracing projects. Additionally, he's the co-host and producer of On-Call Me Maybe, the world's best-named podcast about building and operating reliable software. You can find him on Twitter @austinlparker A big thanks to the event's host, Uber (https://eng.uber.com), and sponsor, Lightstep (https://lightstep.com)! ** Call for speakers ** If you have learnings and insights you'd like to share, we'd love to hear from you. Let us know if you're interested in speaking at an upcoming event!

    2
  • Building A Tracing System at Dropbox and Deriving Insights from Traces

    We're very excited to announce our third Distributed Tracing - NYC meetup at the Uber NYC office! Pizza, beer and wine will be available at 6:20 p.m. and the talks will begin at 7:00 p.m. First talk: Building a Tracing System at Dropbox Presenter: Ross Delinger Distributed tracing is a challenge that is both technical and organizational. In this talk, we’ll explore how Dropbox is solving these challenges. We’ll also cover how our tracing infrastructure has evolved incrementally to keep pace with the migrations and architecture changes required by Dropbox’s move to a service-oriented architecture. Ross is an SRE at Dropbox and leads the Reliability-Frameworks team, which is building a distributed tracing solution. Prior to that Ross worked on a variety of observability and service infrastructure tooling on the Traffic team at Dropbox. Second talk: Gaining Insights from Distributed Traces Presenter: Joe Farro In this talk, we'll explore several techniques for deriving value from distributed traces. The techniques discussed will be human-centric; our objective is to gain insight into our software system and to facilitate decision-making. We'll cover the strengths and weaknesses of each approach, and how they relate to one another. We'll finish with a look at several emerging directions and how they relate to the techniques covered. Joe is a software engineer and member of the Observability team at Uber. He's a core contributor to the Jaeger and OpenTracing CNCF projects. A big thanks to the event's sponsor and host (and my employer), Uber! *Call for speakers* If you have learnings and insights you'd like to share, we'd love to hear from you. Let us know if you're interested in speaking at an upcoming event!

    1
  • Span vs Event: Facebook's and Lightstep's Distributed Tracing Systems @ Uber NYC

    We're very excited to announce our second Distributed Tracing - NYC meetup! Pizza, beer and wine will be available at 6:20 p.m. and the talks will begin at 7:00 p.m. First talk: Tracing and Trace Processing At Facebook Presenters: Edison Gao and Michael Bevilacqua-Linn In this talk, we’ll give a brief overview of Facebook’s distributed tracing system, Canopy, including how and why it differs from the more standard OpenTracing/Dapper model. Then we’ll take a deeper look at one of the more unique features – our streaming trace processing system, which processes streams of traces into aggregate data used for a variety of performance and operational analysis. Edison is a Software Engineer at Facebook on the Canopy team, Facebook's distributed tracing framework. Prior to joining Facebook, he interned at Palantir. Edison graduated from Carnegie Mellon University with a BS in Mathematics and a BS in Computer Science. Michael is a Software Engineer at Facebook on the Canopy team. Previously, he worked at Comcast on a variety of big data, tracing, ip video, and other systems. Second talk: Theory vs. Praxis in Distributed Tracing - The Science of a Span Presenter: Austin Parker What is a span, really? The definition is simple, but sometimes simplicity breeds the greatest complexities. In this talk, you’ll learn some practical lessons based on our experience building monitoring and instrumentation at LightStep. This talk should help you better understand effective instrumentation practices that will lead to more thoughtful, and useful, observability of a system. Austin is an Open Source Software Engineer at LightStep working on OpenTracing. Previously, he was a Software Architect at Apprenda working on a variety of projects including monitoring, observability, and DevOps. You can follow him on Twitter https://twitter.com/austinlparker A big thanks to the event's sponsor and host (and my employer), Uber! Call for speakers We intend to schedule a meetup every six to eight weeks and would love to schedule speakers well beyond that timeframe. Let us know if you're interested in speaking at an upcoming event!

    3
  • Distributed Tracing Meetup

    Uber

    **Our Inaugural Event!** We're very excited to announce our first meetup! Beer and wine will be available at 6:30 p.m. and the talks will begin at 6:45 p.m. Won Jun Jang: Sampling Strategies Most distributed tracing systems must use sampling to keep the amount of traces that their system generates in check. This talk will go over the various sampling strategies available and how we use them at Uber to provide statistically significant insight into the system. Won Jun Jang is a software engineer at Uber on the Observability team where he works primarily on distributing tracing. What he does in his free time is confidential. Mahmoud Saada: Tracing a Kafka Pipeline Kafka pipelines, like most distributed systems, can be hard to debug. Mahmoud, site reliability engineer at Agolo, will show how to instrument Java microservices that communicate over Kafka. This will include the different challenges, and gotchas faced. The talk is suitable for beginners and will include a hands-on demo of instrumenting and debugging a Kafka pipeline using Opentracing, Jaeger, and open source tracing wrappers for datastores and message queues. Lightning Talks We're interested in closing the event with two five-minute lightning talks on distributed tracing and / or Observability. Let us know if you'd like to share your perspective and insights! Thanks to our sponsor! A big thanks to the event's sponsor and host (and my employer), Uber! Call for speakers We intend to schedule a meetup every six to eight weeks and would love to schedule speakers well beyond that timeframe. Let us know if you're interested in speaking at an upcoming event!

    9