O slideshow foi denunciado.
Utilizamos seu perfil e dados de atividades no LinkedIn para personalizar e exibir anúncios mais relevantes. Altere suas preferências de anúncios quando desejar.
Distributed Tracing
Latency analysis for microservices
Reshmi Krishna
@reshmi9k
About Me
 Software Engineer
 Senior Platform Architect, Pivotal
 Conference Speaker
MeetUp : Cloud-Native-New-York
@res...
Agenda
 Distributed Tracing
 Tracers and Tracing Systems
 Zipkin
 Demo – Spring Cloud Sleuth, Zipkin, PCF Metrics
Everything is going to be okay!
Until
Let’s Debug
It doesn’t look like this
QueryHandlerService
IndexerService
BackendService
PageRankingService
Web Frontend
More like this
Troubleshooting Latency issues
 When was the event? How long did it take?
 How do I know it was slow?
 Why did it take ...
Distributed Tracing
 Distributed Tracing is a process of collecting end-to-end transaction graphs in near real
time
 A t...
Tracers
 Tracers add logic to create unique trace ID
 Trace ID is generated when the first request is made
 Span ID is ...
Visualization - Traces & Spans
service1
Trace Id : 1, Span Id : 1
service4
Trace Id : 1, Parent Id : 2, Span Id : 4
servic...
Dapper Paper By Google
@reshmi9k
@reshmi9k
This paper described Dapper, which is Google’s production distributed systems
t...
Zipkin
Zipkin is a distributed tracing system
Implementation based on Dapper paper, Google
Aggregate spans into trace t...
Initial Zipkin Architecture
Demo : Architecture Diagram
Spring Cloud
Sleuth
Collector
Span
Store
Transport
Mq/Http/Log
Spring Cloud
Sleuth
Spring Clou...
Let’s look at some code & Demo
Links
 Dapper, Google : http://research.google.com/pubs/pub36356.html
 Code for this presentation : https://github.com/r...
Distributed tracing
Distributed tracing
Distributed tracing
Distributed tracing
Próximos SlideShares
Carregando em…5
×

0

Compartilhar

Baixar para ler offline

Distributed tracing

Baixar para ler offline

gives rise to a range of benefits including individual scaling and individual deployments. However, it also introduces challenges regarding configuration management, load balancing, and latency analysis. Reshmi Krishna discusses how companies like Twitter analyze microservices latency in real time and demonstrates how to integrate popular distributed tracing tools like Zipkin into an existing application with just a few lines of code. At the end, we will also see a demo of tracing capabilities from PCF Metrics.

  • Seja a primeira pessoa a gostar disto

Distributed tracing

  1. 1. Distributed Tracing Latency analysis for microservices Reshmi Krishna @reshmi9k
  2. 2. About Me  Software Engineer  Senior Platform Architect, Pivotal  Conference Speaker MeetUp : Cloud-Native-New-York @reshmi9k
  3. 3. Agenda  Distributed Tracing  Tracers and Tracing Systems  Zipkin  Demo – Spring Cloud Sleuth, Zipkin, PCF Metrics
  4. 4. Everything is going to be okay!
  5. 5. Until
  6. 6. Let’s Debug
  7. 7. It doesn’t look like this QueryHandlerService IndexerService BackendService PageRankingService Web Frontend
  8. 8. More like this
  9. 9. Troubleshooting Latency issues  When was the event? How long did it take?  How do I know it was slow?  Why did it take so long?  Which microservice was responsible?
  10. 10. Distributed Tracing  Distributed Tracing is a process of collecting end-to-end transaction graphs in near real time  A trace represents the entire journey of a request  A span represents single operation call  Distributed Tracing Systems are often used for this purpose. Zipkin is an example  As a request is flowing from one microservice to another, tracers add logic to create unique trace Id, span Id
  11. 11. Tracers  Tracers add logic to create unique trace ID  Trace ID is generated when the first request is made  Span ID is generated as the request arrives at each microservice  Example tracer is Spring Cloud Sleuth  Tracers execute in your production apps! They are written to not log too much  Tracers have instrumentation or sampling policy
  12. 12. Visualization - Traces & Spans service1 Trace Id : 1, Span Id : 1 service4 Trace Id : 1, Parent Id : 2, Span Id : 4 service2 Trace Id : 1, Parent Id : 1, Span Id : 2 service3 Trace Id : 1, Parent Id : 2, Span Id : 3
  13. 13. Dapper Paper By Google @reshmi9k @reshmi9k This paper described Dapper, which is Google’s production distributed systems tracing infrastructure Design Goals : Low overhead Application-level transparency Scalability
  14. 14. Zipkin Zipkin is a distributed tracing system Implementation based on Dapper paper, Google Aggregate spans into trace trees Manages both collection and lookup of the data In 2015, OpenZipkin became the primary fork
  15. 15. Initial Zipkin Architecture
  16. 16. Demo : Architecture Diagram Spring Cloud Sleuth Collector Span Store Transport Mq/Http/Log Spring Cloud Sleuth Spring Cloud Sleuth Spring Cloud Sleuth Query ServerZipkin UI ZIPKIN APP APP APP APP
  17. 17. Let’s look at some code & Demo
  18. 18. Links  Dapper, Google : http://research.google.com/pubs/pub36356.html  Code for this presentation : https://github.com/reshmik/DistributedTracingDemo_Velocity2016.git  Sleuth’s documentation: http://cloud.spring.io/spring-cloud-sleuth/spring-cloud-sleuth.html  Repo with Spring Boot Zipkin server: https://github.com/openzipkin/zipkin-java  Zipkin deployed as an PCF https://github.com/spring-cloud-samples/sleuth-documentation- apps/tree/master/zipkin-server  Pivotal Web Services trial : https://run.pivotal.io/  PivotalCloudFoundry on your laptop : https://docs.pivotal.io/pcf-dev/ @reshmi9k

gives rise to a range of benefits including individual scaling and individual deployments. However, it also introduces challenges regarding configuration management, load balancing, and latency analysis. Reshmi Krishna discusses how companies like Twitter analyze microservices latency in real time and demonstrates how to integrate popular distributed tracing tools like Zipkin into an existing application with just a few lines of code. At the end, we will also see a demo of tracing capabilities from PCF Metrics.

Vistos

Vistos totais

655

No Slideshare

0

De incorporações

0

Número de incorporações

23

Ações

Baixados

20

Compartilhados

0

Comentários

0

Curtir

0

×