Why Fast Matters

Why Fast Matters

We’re all—I hope!—well aware that performance is important; it’s great for business and it’s great for our users. But things are still not fast enough. With more and more emerging markets coming online, and more and more apps moving to the web platform, we’re reaching an intersection where connections are getting slower and websites are getting heavier. In this talk, we’ll learn just what these emerging markets mean to us, and how we can begin to move in the right direction.

Bb854891c46db72f4a6f9da4504e879a?s=128

Harry Roberts

May 17, 2017
Tweet

Transcript

  1. WHY FAST MATTERS Harry Roberts | infoShare, Gdańsk | May

    2017
  2. HI, I’M HARRY Consultant Performance Engineer Leeds, UK @csswizardry csswizardry.com

  3. None
  4. THE TRAINLINE REDUCED LATENCY BY 0.3 SECONDS […] AND CUSTOMERS

    SPENT AN EXTRA £8.1M A YEAR
  5. NETFLIX SAW A 43% DECREASE IN THEIR BANDWIDTH BILL AFTER

    TURNING ON GZIP
  6. GQ CUT LOAD TIME BY 80% AND SAW AN 80%

    INCREASE IN TRAFFIC. MEDIAN TIME SPENT ON THE SITE INCREASED BY 32%
  7. WPOSTATS.COM

  8. 1. IT WILL MAKE YOU MONEY 2. IT WILL SAVE

    YOU MONEY 3. IT MAKES USERS HAPPIER
  9. IT’S NOT JUST FINANCIAL…

  10. — My Buddy Warren “Sorry I didn’t reply to your

    email mate… I could see it but couldn’t open it because the internet out [in Thailand] is s—t.”
  11. — Anon “I am currently at my parents’ place in

    Rawatbhata, Rajasthan [India]. Since my parents don’t have a computer they only consume internet through their smartphone […] providers which in our town are still 2G. Right now I have connected my laptop via WiFi hotspot. Opening Gmail in basic HTML version takes 30s to a minute.”
  12. None
  13. THE NEXT BILLION USERS

  14. qz.com/on/the-next-billion

  15. csswz.it/2pa7bX0

  16. BANGLADESH ! Average connection: 3,634kbps† 14.4% of the country is

    online* 3.9m broadband subscriptions (2.4 per 100 people)* 133.7m cellular subscriptions (83.4 per 100 people)*
  17. 34× MORE MOBILE THAN BROADBAND

  18. INDIA " Average connection: 3,465kbps† 26% of the country is

    online* 17.1m broadband subscriptions (1.3 per 100 people)* 1b cellular subscriptions (78.8 per 100 people)*
  19. 58× MORE MOBILE THAN BROADBAND

  20. PAKISTAN # Average connection: 2,508kbps† 18% of the country is

    online* 1.8m broadband subscriptions (1 per 100 people)* 125.9m cellular subscriptions (66.9 per 100 people)*
  21. 70× MORE MOBILE THAN BROADBAND

  22. INDONESIA $ Average connection: 4,543kbps† 22% of the country is

    online* 2.8m broadband subscriptions (1.1 per 100 people)* 338.4m cellular subscriptions (132.3 per 100 people)*
  23. 121× MORE MOBILE THAN BROADBAND

  24. ! " # $ 3,537.5KBPS AVERAGE SPEED 20.1% OF PEOPLE

    ARE ONLINE 1.45% HAVE BROADBAND 90.35% ON CELLULAR
  25. % 14,936KBPS AVERAGE SPEED 92% OF PEOPLE ARE ONLINE 37.7%

    HAVE BROADBAND 125.8% ON CELLULAR
  26. & 12,754KBPS AVERAGE SPEED 68.0% OF PEOPLE ARE ONLINE 19.5%

    HAVE BROADBAND 148.7% ON CELLULAR
  27. WHAT DOES THIS TELL US?

  28. WE’RE BUILDING FOR A TOTALLY DIFFERENT PROFILE OF USER

  29. HOW FAST IS FAST ENOUGH?

  30. HOW FAST IS FAST ENOUGH? It’s hard to say… You

    can run benchmarking Get a feel for things over time
  31. JUST BE FASTER THAN YOUR NEAREST COMPETITOR…

  32. dareboost.com

  33. speedcurve.com

  34. GETTING THERE

  35. STEP 0: WANT A FAST WEBSITE

  36. WANT A FAST WEBSITE Do not underestimate the importance of

    this step Things get markedly easier if you actually try Make sure it’s a business goal/priority Ensure that people are putting effort into performance
  37. STEP 1: UNDERSTAND THE PROBLEM

  38. …PROPERLY UNDERSTAND IT

  39. None
  40. charlesproxy.com

  41. IT’S NOT JUST CONNECTION SPEED ANYMORE…

  42. None
  43. None
  44. geekaphone.com

  45. csswz.it/2pL4MlZ

  46. None
  47. THERE’S NO REPLACEMENT FOR REAL DEVICES

  48. None
  49. None
  50. BUILD UP AN IDEA OF REALISTIC CONDITIONS

  51. STEP 2: KNOW WHAT’S GOING ON

  52. WHAT DOES THIS SCRIPT DO? WHICH TEAM IS IN CHARGE

    OF THIS THING? ARE WE EVEN USING THIS?
  53. KNOW WHAT’S GOING ON Other people/teams add things to your

    site all the time Tag managers, social widgets, tracking scripts, analytics Schrödinger effect: measuring/optimising has its own overhead
  54. CALL MEETINGS Work out what everything is Who is using

    it Do they need it What can you do about it
  55. None
  56. csswz.it/2qdCJzg

  57. KNOW YOUR LIABILITIES…

  58. None
  59. Performance » Bottom-Up » Group By Domain

  60. None
  61. IDENTIFYING THIRD PARTIES

  62. None
  63. None
  64. None
  65. BEING VULNERABLE

  66. None
  67. None
  68. ## # SPoF Spoofing # # Point third-party domains at

    WPT’s blackhole server. ## 72.66.115.13 platform.twitter.com 72.66.115.13 connect.facebook.net 72.66.115.13 fonts.googleapis.com 72.66.115.13 assets.adobedtm.com 72.66.115.13 fast.fonts.net
  69. DON’T PRIORITISE YOUR OWN METRICS OVER YOUR USERS’ EXPERIENCES

  70. STEP 3: MEASURE EVERYTHING

  71. HOW DO WE KNOW:
 1) WHAT’S WRONG? 
 2) WHEN

    IT’S RIGHT?
  72. None
  73. None
  74. None
  75. None
  76. jana.com/insights

  77. None
  78. None
  79. None
  80. None
  81. BUDGETING

  82. PERFORMANCE BUDGETS Basically just monitoring with alerts After a while,

    you’ll begin to know how fast you should be Set budgets and tolerances Endeavour to stay within those limits
  83. None
  84. Not the most complex site in the world But fairly

    typical Plenty going on CSSWIZARDRY.COM
  85. None
  86. None
  87. None
  88. None
  89. None
  90. None
  91. None
  92. CLOSING…

  93. CLOSING Care: Actually prioritise, consider, and champion performance Understand: Your

    customers, the problem, the landscape Measure: Everything you can, before and after
  94. STATISTICS AND DATA Tim Kadlec, Tammy Everts * World DataBank

    † Akamai State of the Internet ‡ ITU Measuring the Information Society Report
  95. THANK YOU Harry Roberts @csswizardry csswizardry@gmail.com speakerdeck.com/csswizardry