Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Predictive Prefetching for the Web

82bafb0432ce4ccc9dcc26f94d5fe5bc?s=47 Minko Gechev
November 11, 2019
Programming
0
36

Predictive Prefetching for the Web

82bafb0432ce4ccc9dcc26f94d5fe5bc?s=128

Minko Gechev

November 11, 2019
Tweet

More Decks by Minko Gechev

See All by Minko Gechev
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
0
15
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
0
40
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
1
68
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
1
160
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
1
28
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
1
61
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
2
710
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
1
84
82bafb0432ce4ccc9dcc26f94d5fe5bc?s=48 mgechev
2
2.3k

Other Decks in Programming

See All in Programming
Bbed8642e7e79c1e4c6ea89aa9934407?s=48 elijo
2
110
3931098ae486b6df619050c63e24f6cc?s=48 osyo
1
220
028088964f255d677d471558da865b72?s=48 satoshin
2
170
8f84b7d8869ef6005d89b378e8661f7c?s=48 masuda220
0
120
Afdb8e40066f7939ff82727bc0c892ce?s=48 junk0612
0
120
8e844e43f029ec38e206a464cf4587b3?s=48 ttskch
14
15k
3931098ae486b6df619050c63e24f6cc?s=48 osyo
0
820
9bda8755a689204677a198ee85c0eb07?s=48 banjun
PRO
0
190
Af64bc38c0ffcfcabdf430759ee491ce?s=48 lovee
1
410
06641ec93c5914e0db74cddff2ec6004?s=48 objectiveaudio
0
180
23a9ea59992e5349aa4bf659a6d583d3?s=48 nanashiki
1
560
15934fa2aa7b2ce21f091e9b7cffa856?s=48 manfredsteyer
PRO
0
120

Featured

See All Featured
027d1ebf66cc039a0bd3b55eeadbe75d?s=48 sachag
448
36k
78b475797a14c84799063c7cd073962f?s=48 holman
464
280k
5f83ef806155b403c3393160ce51f955?s=48 jlugia
218
16k
96270e4c3e5e9806cf7245475c00b275?s=48 addyosmani
1352
200k
Fde6c3a50ca518a909ef35c22c0ee0e4?s=48 destraynor
148
20k
8081b26e05bb4354f7d65ffc34cbbd67?s=48 chriscoyier
147
20k
5f2da528927a2ec9ba4fec2069cbc958?s=48 kneath
297
39k
2bb923c57a1fdc3a2eb484bb8d565fd2?s=48 ufuk
60
7.2k
2f5463832ccb768ccb4a1ca3607c27ef?s=48 jacobian
260
20k
E195ae45320d9202eaa01c9f1d31a416?s=48 marktimemedia
14
660
24fc194843a71f10949be18d5a692682?s=48 gr2m
86
11k
64827b31aef81498662e8af4bd6d5157?s=48 jonrohan
1021
400k

Transcript

  1. None

  2. Minko Gechev @mgechev Predictive Prefetching for the Web for better

    user experience
  3. None
  4. None
  5. None
  6. None

  7. Prefetching for speeding up navigation faster Web and fewer rage

    clicks

  8. ⚡ Prefetching Navigation

  9. Prefetching strategies • Prefetch all • Visible links • On

    mouse over • Predictive prefetching

  10. Web app with instant transitions Application usage report Data analytics/ML

    model How predictive prefetching works
  11. None
  12. None

  13. github.com/guess-js

  14. None

  15. Naveed Ahmed

  16. Search and Find Latest Jobs in Pakistan mustakbil.com

  17. Search and Find Latest Jobs in Pakistan mustakbil.com

  18. None

  19. Prefetching data consumption one user at the home page 2.56MB

    Prefetch all modules 100% 1.22MB Prefetch visibile links 48% 0.18MB Predictive prefetching 7%

  20. 90% for users on fast 3G prefetching accuracy >

  21. Additional optimizations implemented by Guess.js Adapts to the connection speed

    Prefetches by likelihood Calculates likelihood at build time Does not block the main thread

  22. Using Guess.js

  23. Predictive Prefetching for the Web Speeds up page navigation Increases

    user satisfaction Support for popular frameworks Reduces data usage

  24. web.dev/link-prefetch web.dev/predictive-prefetching

  25. Thank you! Minko Gechev @mgechev