Data Visualization in the Trenches

Data Visualization in the Trenches

This talk was given at Bocoup's OpenVis Conf in Boston.

6601d82cf1b6776afd9c31f3d18294c3?s=128

Abe Stanway

May 17, 2013
Tweet

Transcript

  1. Abe Stanway @abestanway Data visualization in the trenches

  2. None
  3. 1.5 Billion page views $117 Million of goods sold 950

    thousand users
  4. 1.5 Billion page views $117 Million of goods sold 950

    thousand users (in December)
  5. Text We practice continuous deployment.

  6. de • ploy /diˈploi/ Verb To release your code for

    the world to see, hopefully without breaking the Internet
  7. 250+ committers, everyone deploys.

  8. Day one: Deploy.

  9. None
  10. 30+ DEPLOYS A DAY

  11. Text “30 deploys a day? Is that safe?”

  12. Text Yes, with the proper tooling.

  13. Text Every engineer must have a finger on the pulse

    of the system.
  14. Text How do you make an entire web stack “consumable”

    to a handful of engineers?
  15. Text More information Quickly consumable More abstraction

  16. p

  17. None
  18. Text Real time error logging

  19. Text No abstraction. Fluffy information. Easy to consume.

  20. “Not all things that break throw errors.” - Oscar Wilde

  21. 1. ssh to server 2. poke around for the log

    files 3. try to remember what they mean 4. try to scroll back in time to find when they started acting up. 5. repeat
  22. Text No abstraction. Fluffy information.

  23. Text ...but hard to get at. Lots of friction means

    not easily consumable.
  24. Text Bump up a layer of abstraction.

  25. 1. create a graph 2. look at the graph 3.

    ?? 4. profit!!
  26. StatsD

  27. StatsD::increment(“foo.bar”)

  28. If it moves, graph it!

  29. If it doesn’t move, graph it anyway (it might make

    a run for it)
  30. None
  31. Text Some abstraction. Denser information.

  32. Text ...still hard to get at en masse.

  33. Text Bump up a layer of abstraction.

  34. DASHBOARDS!

  35. Hang out with the dashboards after you push.

  36. None
  37. [1358731200, 20] [1358731200, 20] [1358731200, 20] [1358731200, 20] [1358731200, 20]

    [1358731200, 20] [1358731200, 20] [1358731200, 20] [1358731200, 60]
  38. DASHBOARDS x 250000 !

  39. None
  40. “...but there are also unknown unknowns - there are things

    we do not know we don’t know.”
  41. Text Slightly denser information is negated by the deluge.

  42. Text The majority remains unconsumable.

  43. Text Bump up a layer of abstraction.

  44. SKYLINE

  45. A real time anomaly detection system

  46. None
  47. Text Very abstract, harder to understand, but the effective information

    density is massive.
  48. Text Consumption is also increased by outsourcing it to the

    machine.
  49. Text Trust becomes an issue.

  50. So you found an anomaly.

  51. MAYBE THERE ARE OTHERS

  52. Oculus: a metrics correlation system

  53. None
  54. Text A good tool adds “touch” to the system.

  55. Text More “touch” means more intimacy with the stack.

  56. Text A delicate balance between insight, consumptive capacity, and actionability.

  57. Text More abstraction leads to less intimacy...

  58. Text ...but greater information density.

  59. @abestanway abe@etsy.com Abe Stanway Data Engineer Thanks!