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

Improving Search With One Weird Trick

25e2ecf9b520e06d71e47ab083924300?s=47 xyu
January 26, 2017

Improving Search With One Weird Trick

25e2ecf9b520e06d71e47ab083924300?s=128

xyu

January 26, 2017
Tweet

More Decks by xyu

Other Decks in Technology

Transcript

  1. Xiao Yu / Automattic xyu.io Improving Search With One Weird

    Trick Elasticsearch Meetup 2017
  2. WordPress.com • 22.4 billion views • 73.9 million posts •

    49.0 million comments
  3. WordPress.com + Elasticsearch • 22.4 billion views • 73.9 million

    posts • 49.0 million comments • 16.2 TB primary shards • 4.4 billion docs • 1.3 billion queries (+0.5 billion langdetect)
  4. None
  5. Recommending Content Globally

  6. None
  7. It’s a hackathon, so…

  8. None
  9. None
  10. None
  11. Fake it 'til you make it

  12. None
  13. None
  14. None
  15. None
  16. None
  17. GET /_search?routing= ${USER}${N*DAY}, ${USER}${N*DAY+1}, ${USER}${N*DAY+2}, ${USER}${N*DAY+3},… { “query”:{…} }

  18. S1 S2 S3 S4 S5 S6 S7 S8 S9 S10

    S11 S12 S13 S14 S15 S16 S17 S18 S19 S20 Day 0 Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 8 Day 9
  19. 60 6 ~18 ~33% Shards / Index Skip Multiple (10

    day cycle) Shard Batch (~30% load) Fresh Content / Day
  20. One Weird Trick S1 S2 S3 S4 S5 S6 S7

    S8 S9 S10 S11 S12 S13 S14 S15 S16 S17 S18 S19 S20 Day 0 Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 8 Day 9
  21. automattic.com/work-with-us/search-wrangler/

  22. Xiao Yu / Automattic xyu.io Thanks! Elasticsearch Meetup 2017