Successfully reported this slideshow.
Your SlideShare is downloading. ×

How to Deliver a Critical and Actionable Customer-Facing Metrics Product with InfluxDB | Cullen Murphy | Particle

Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad

Check these out next

1 of 34 Ad

How to Deliver a Critical and Actionable Customer-Facing Metrics Product with InfluxDB | Cullen Murphy | Particle

Download to read offline

In IoT, understanding the health of thousands of devices is critical for deployment at scale especially when troubleshooting an issue. Particle’s customer base needed visibility into their devices with actionable data to reference in real time. Join Cullen Murphy, Site Reliability Engineer at Particle, to learn how the team built a metrics system on Telegraf, Kubernetes, and Prometheus to deploy a customer-facing product that provides critical and relevant data to their IoT product creators.

In IoT, understanding the health of thousands of devices is critical for deployment at scale especially when troubleshooting an issue. Particle’s customer base needed visibility into their devices with actionable data to reference in real time. Join Cullen Murphy, Site Reliability Engineer at Particle, to learn how the team built a metrics system on Telegraf, Kubernetes, and Prometheus to deploy a customer-facing product that provides critical and relevant data to their IoT product creators.

Advertisement
Advertisement

More Related Content

Slideshows for you (20)

Similar to How to Deliver a Critical and Actionable Customer-Facing Metrics Product with InfluxDB | Cullen Murphy | Particle (20)

Advertisement

More from InfluxData (20)

Recently uploaded (20)

Advertisement

How to Deliver a Critical and Actionable Customer-Facing Metrics Product with InfluxDB | Cullen Murphy | Particle

  1. 1. How to deliver customer-facing metrics with InfluxDB
  2. 2. Cullen Murphy - Site Reliability Engineer - Particle ● Originally started in Embedded Systems ● Automated billing systems with Ruby on Rails ● Previously a Contracting WebDev and SysAdmin ● At Particle for 2 years
  3. 3. What is Particle?
  4. 4. WE HAVE GROWN THE WORLD’S LARGEST IOT DEVELOPER COMMUNITY 170,000
  5. 5. WE FOCUS ON SOLVING REAL PROBLEMS FOR REAL CUSTOMERS
  6. 6. & Particle Internal Monitoring ● Kubernetes DaemonSet ● StatsD DataDog tagging ● Application Specific Metrics Fleet Health Metrics ● Telegraf Sidecar ● Prometheus Client ● Device Specific Metrics
  7. 7. WHY BUILD IT HOW WE BUILT IT HARD PROBLEMS
  8. 8. WHY BUILD IT HOW WE BUILT IT HARD PROBLEMS
  9. 9. InfluxDB provides deep insight Why can’t our customers have this?
  10. 10. Cloud Computing vs IoT ● Has some CPUs ● Has a bunch of RAM ● Runs some software ● Connected over a network ● Has limited resources ● Usually 10s or 100s in a deployment Cloud Computer IoT “thing” ● Has a CPU ● Has some RAM ● Runs some software ● Connected over a network ● Has limited resources ● Usually 1,000s or 100,000s in a deployment
  11. 11. NO METRICS POSTERS Metrics without actionable data are useless
  12. 12. So what did we build?
  13. 13. THINK ABOUT DATA SECURITY Don’t store sensitive data THINK ABOUT DATA SECURITY Don’t store sensitive data
  14. 14. Architecture
  15. 15. HOW WE BUILT IT WHY BUILD IT HARD PROBLEMS
  16. 16. EASY COLLECTION From a development standpoint but might make storage hard Continuous Queries (or Scheduled Tasks) can help
  17. 17. EASY COLLECTION with a NPM Module Setup Use
  18. 18. EASY COLLECTION with a NPM Module Setup Use
  19. 19. COMPLEXITY IN STORAGE QUERIES Continuous Queries work but... ● InfluxQL is a bit limiting ● InfluxData’s Support has been fantastic
  20. 20. COMPLEXITY IN STORAGE QUERIES Continuous Queries work but... ● InfluxQL is a bit limiting ● InfluxData’s Support has been fantastic ● Flux is far more concise for our use case
  21. 21. WHY BUILD IT HOW TO BUILD IT HARD PROBLEMS
  22. 22. LYING METRICS Are worse than no metrics If you can’t say something nice true, don’t say anything at all
  23. 23. Collect only what you need
  24. 24. PROVIDE GUARD RAILS Providing data access is powerful, but keep a tight ship
  25. 25. PROVIDE GUARD RAILS Providing data access is powerful, but keep a tight ship PROVIDE GUARD RAILS Providing data access is powerful, but keep a tight ship
  26. 26. THINK ABOUT ACCESS Ensure your customers can only access their data
  27. 27. THINK ABOUT DATA SECURITY Don’t store sensitive data
  28. 28. WHY BUILD IT HOW WE BUILT IT HARD PROBLEMS
  29. 29. THINK ABOUT DATA SECURITY Don’t store sensitive data THINK ABOUT DATA SECURITY Don’t store sensitive data
  30. 30. QUESTIONS?
  31. 31. We’re Hiring! particle.io/jobs
  32. 32. Architecture

×