Spotify Pitch Deck

Spotify Pitch Deck

1. Spotify,

Distributed Systems

Jon Áslund
@jooon

2. storage

consistent hashing

3. playlist

4. playlist

cassandra

5. playlist

structured key-value
store

6. playlist

high throughput

7. playlist

choose redundancy and

consistency level

8. playlist

datacentre aware

9. p2p

10. Data source ratio by week

Weekend
Weekday Bar
Morning

11. thank you

12. Master Thesis?

13. Yes, please!

14. or come and work for us

15. Oh, and
we have
cake as

well! :D

spotify.com/jobs

16. Who am I?

17. Jon Aslund

19. Computer Science

20. GD

TR,
ÉKTHS

VETENSKAP
39 OCH KONST 9 る

a

21. Shakespeare

Programming
Language

22. Joined Spotify

Autumn 2006

23. April 2006

the idea

26. “music itself is going to become
like running water or electricity”

David Bowie, 2002

31. February 2012

32. 10+ million users

(active within previous 30 days)

33. 15+ million tracks

34. 500+ million playlists

35. 1000+ servers

in 3 countries

36. Spotify is a distributed

system

37. Fast
<200ms

38. Always working

39. How do you

accomplish that?

40. Cheat

as much as possible

41. Fast

vs
Feels fast

42. Always working
vs

Pretend there is
nothing wrong

43. Build to fail

44. Redundancy

(talk about waffles)

45. Error recovery

46. Don't forget to test

your recovery code

47. Fire drill: Cut the

atlantic cable

48. Netflix

Chaos Monkey

49. “On January 15, 1990,
AT&T’s long-distance

telephone switching system
crashed.”

50. Isolation

51. Getting the right tool
for the job

52. certify | 19 op py \ Bosse - Fin (Ob 3.15

5046 I Er

M UTE IM0DER-
PL1GEK つづ し 10 -
TEKET

Latar & filer
=

WER
BEIM

53. Master Storage

5 j Playlist

54. Consistency

Availability
Partition tolerance

http://ksat.me/a-plain-english-introduction-to-cap-theorem/

55. Availability

Partition
Tolerance

56. user

57. PostgreSQL

59. search

60. search

Lucene index in

memory

61. replaced daily