MongoDB DC 2012: Mongo or Die - How MongoDB Powers "Doodle or Die"

Aaron Silverman, Doodle or Die
"Doodle or Die" is a popular online drawing game built on Node.js and MongoDB. It started off as an entry to the 2011 Node Knockout competition and after winning the category for "Most Fun" has continued to grow into a game that has thousands of players play each day who have produced millions of drawings. This talk will use Doodle or Die as a vehicle to showcase the many strengths of using MongoDB as a go-to database for small and midsize applications. It will also cover lessons learned as we used MongoDB to rapidly develop and scale Doodle or Die.

Transcript

1. Doodle  or  Die   @DoodleOrDie  
2. What  is  Doodle  or  Die?  
3. Telephone   I'd  like  some  beer!   I'd  like  some  deer   I  see  no  deer   I’ve  no  idea!   Phrase   Phrase   Phrase   Phrase  
4. Doodle  or  Die   Shining  Apple   Doodle   Phrase   Phrase   Eat  your  fruit  or  DIE!   Doodle  
10. What  Powers  Doodle  or  Die?  
12. Started  very,  very  small         4  Cores   128MB  RAM   Node  Server   MongoDB  Server  
13. Small  -­‐  2GB   Got  serious  about  our  servers   MongoDB  Server         8  Cores   256MB  RAM   Node  Server  
14. Large  -­‐  5GB   Called  in  some  Reinforcements   MongoDB  Server         12  Cores   1GB  RAM   Node  Server  
15. In  the  last  30  Days:   • 2,500,000  page  views   • 100,000  uniques   • 35,000  acSve  player  accounts     • 2,000,000  new  doodles  and  descripSons  
16. “Small  Data”   Mongo  DB   4  GB  data  size   <1  GB  index  size   ~10  queries/sec     Amazon   170  GB  data  size   8  GB  in/month   200  GB  out/month     •  Player  Info   •  Chain  Info  (excluding   doodles)   •  Group  Info   •  Game  State   •  Logs   •  Doodles   •  StaSc  Content   •  Compressed   Database  Backups  
17. MongoDB  -­‐  $65  /  month   Daily  backups  to  Amazon  S3:  $16/mo  
18. Amazon  S3  -­‐  $70  /  month  
19. Node  -­‐  $62/  month  
20. Total  Cost  To  Host:  $197/month   MongoDB   $65   Amazon  S3   $70   Node     $62   Total   $197  
21. PAAS  Provides  Easy  Upgrade  Path  
22. General  Principles  
23. Custom  _id   ParSally  random  string  generated  using   ShortId  node  module   ObjectId   ObjectId("4fd02d5d78315a502d15cdde")   ObjectId("4fd02d5a78315a502d15cddd")   ObjectId("4fd02d5878315a502d15cddc")   ShortId   "8rOIwh2VD"   "1qyY61Lu1"   "5GQnbx-­‐1"  
24. •  Shorter,  less  cumbersome  in  code  and   queries   db.players.findOne({_id: ‘58mwYlTKV’});! db.chains.update({_id: ‘58mwYlTKV’}, ! {$set: activePlayer_id: ‘88ueYaL6V’});! hfp://doodleordie.com/c/5ONtvvSGH   •  Randomness  could  help  with  sharding;   more  importantly  makes  it  harder  to   cheat     <span class="doodle" data-jsonp="http:// doodles.s3.amazonaws.com/d2/Eh8-Po2R5/1Em5kj3LY.js">!
25. Ques]on  Oriented  Subdocuments   What  chain  is  this  player  working  on  right  now?   What  are  this  player’s  stats?   Which  players  are  not  eligible  to  be  assigned  this  chain?  
26. Goal  is  for  most  “QuesSons”  to  be  able  to   be  answered  in  one  query  from  one  sub   document     Related  “QuesSons”  will  share  common   ancestors     db.players.findOne({_id: ‘58mwYlTKV’}, ! {‘game.recentSkips’: 1});! db.players.findOne({_id: ‘58mwYlTKV’}, ! {game: 1});!
27. Indexes  are  designed  to  make  answering   quesSons  easy!     Which  player  is  working  on  this  chain?   db.players.ensureIndex({‘game.activeChain_id’: 1});! What  chains  are  recently  awaiSng  a  new  doodle?   db.chains.ensureIndex({inUse: -1, ! activeState: 1! lastModified" : -1});!
28. Doodle  or  Die  Collec]ons  
29. Disclaimer:  Much  of  the  original  Doodle  or   Die  code  and  schema  were  created  during  a   weekend  long  hackathon!  
31. players   chains   groups   sessions   log   Primary   Support  
32. Players  
33. players   account   info   login   chainHistory   game   stats   Query   Frequency   Onen   Rarely  
34. players   account   info   login   chainHistory   stats   game     •  acSveState   •  acSveChain_id   •  acSveStepIndex   •  recentSkips   Answers  quesSon:    “What  is  this  player  working  on  right  now?”   db.players.findOne({_id: ‘58mwYlTKV’}, {game: 1});!
35. players   account   info   login   stats   game   chainHistory   Chain1   •  datePlayed   •  dateViewed       ChainN       Answers  the  quesSon:    “What  has  the  player  worked  on?”   db.players.findOne({_id: ‘58mwYlTKV’}, {chainHistory: 1});!
36. players   account   info   login   stats   game   chainHistory   Chain1   •  datePlayed   •  dateViewed       ChainN       YUCK!   Plan  to  refactor  out  along  with  a  refactor  of   how  chains  store  steps  
37. players   account   info   login   game   chainHistory   Answers  the  quesSon:    “How  acSve/good  is  this  player?”   stats   •  totalSteps   •  drawSteps   •  phraseSteps   •  numSkips   •  numLikes   db.players.findOne({_id: ‘58mwYlTKV’}, {stats: 1});!
38. Chains  
39. chains     •  acSvePlayer_id   •  acSveState   •  numSteps   •  lastModified       ineligiblePlayer_ids   steps   chains  are  assigned  (not  as  many  quesSons)   db.chains.findOne({inUse: false,! activeState : player.game.activeState,! ineligiblePlayer_ids: {$ne: player._id},! lastModified: { $gte: timeRange}});!
40. chains     •  acSvePlayer_id   •  acSveState   •  numSteps   •  lastModified     *   steps   ineligiblePlayer_ids   [player_id1    player_id2,    player_id3                            …    player_idN]   Note:  $addToSet  and  $pull  work  great  in   maintaining  this  array  
41. chains     •  acSvePlayer_id   •  acSveState   •  numSteps   •  lastModified     *   ineligiblePlayer_ids   steps   [                     ]     Step1   •  player_id   •  state   •  date       StepN       content  
42. •  phrase     Descrip]on  Step  Content:   Doodle  Step  Content:   •  url  (points  to  S3)   •  Sme   •  numStrokes   •  numDisSnctColors  
43. •  phrase     Descrip]on  Step  Content:   Doodle  Step  Content:   •  url  (points  to  S3)   •  Sme   •  numStrokes   •  numDisSnctColors  
44. doodles   •  player_id   •  date   •  url  (points  to  S3)   •  Sme   •  numStrokes   •  numDisSnctColors   descripSons   •  player_id   •  date   •  text       We  plan  to  stop  embedding  steps  in  chains,   and  link  to  “doodles”  and  “descripSons”   collecSons  
45. How  Does  it  all  Work?  
47. players  queried  for  users  who  are  associated   with  the  authenScated  twifer  account   chain  history  used  to  load  thumbnails,   previous  chain  is  loaded   db.players.find({‘login.twitter.uid’: ‘XXXXXXX’},! ! ! ! ! {game: 1, chainHistory: 1});! db.chains.find({_id: {$in: [player.game.activeChain_id,! player.game.lastChain_id]}});!
49. chains  collecSon  searched  and  atomically   updated  for  unclaimed  eligible  chain  to  be   given  to  player   db.chains.findAndModify(! {inUse: false,! activeState : player.game.activeState,! ineligiblePlayer_ids: {$ne: player._id},! lastModified: { $gte: timeRange}},! {$set: {inUse: true,! activePlayer_id: player._id,! $addToSet: ! {ineligiblePlayer_ids: player._id}});!
51. Content  saved  to  chain   db.chains.update({_id: chain._id, ! activePlayer_id: player._id}},! {$inc: {numSteps: 1},! $set: inUse: false,! lastModified: datePlayed}! $unset: {activePlayer_id: 1},! $push: {steps: {! player_id: player._id,! state: chain.activeState,! content: content,! date: datePlayed}}});!
53. Doodle  strokes  will  be  saved  to  S3   (but  url  to  S3  and  metadata  saved  to  chain)   {"color":"#000000","size":15,"path": [307,66,308,66,308,68,308,69,308,70,308,71,308,72,308,73,306,76,305,79,305,81,302, 83,302,84,302,85,302,86,301,87,300,89,300,90,300,91,300,92,300,95,300,97,300,102,3 00,103,300,104,300,108,300,109,300,110,301,114,303,116,304,116,305,119,305,121,3 07,124,309,127,310,130,310,131,313,132,314,133,316,137,317,138,320,140,321,140,3 23,143,326,143,328,143,333,144,337,144,341,144,343,144,347,143,352,143,354,141,3 57,140,358,140,359,139,362,138,363,138,365,137,368,135,369,132,370,131,371,130,3 74,128,375,128,376,125,378,125,379,124,380,123,381,123,382,120,385,119,386,118,3 88,115,391,112,394,109,394,107,394,106,397,104,397,103,397,102,397,101,397,100,3 97,98,397,96,397,94,397,93,397,91,397,90,397,88,397,87,397,86,397,83,395,80,395,79 ,395,77,394,74,393,72,393,70,393,67,392,65,391,63,389,62,388,59,386,57,383,54,383, 52,381,49,379,48,378,47,376,46,375,44,374,43,372,43,370,42,369,42,368,42,364,41,36 2,41,359,41,355,41,351,42,349,42,347,42,346,44,343,44,342,45,339,46,337,47,336,48, 333,49,332,51,330,51,328,51,327,52,326,53,323,53,322,53,321,54,320,55,317,55,316,5 6,314,58,309,59,306,61,306,62,305,62,304,63]}!
55. Player’s  chainHistory  retrieved   db.chains.find({$in: chain_idArr});! ! db.players.find({‘login.urlSafe’: urlSafeUid},! {chainHistory: 1});! Retrieved  chains  ordered  (on  server)  based   on  previously  sorted  chainHistory   chainHistory  filtered  and  sorted  on  server,   Applicable  chains/steps  retrieved  
57. Stats  are  loaded  from  the  counSng  log  which   is  essenSally  a  bunch  of  counters   incremented   db.log.find({_id: {$in: [‘2012-06-20’, ‘2012-06-20’, ! ! ! ! ! ! ! ‘2012-06-20’, ‘2012-06-20’]}});! Extremely  simple  implementaSon  using   nested  subdocuments  for  organizaSon  
59. Some  Addi]onal  Doodle  or  Die  Tricks  
60. Build  assump]ons  into  queries  
61. Eve  tries  to  join  despite  being  banned!   groups: ! {! _id: ‘8rOIwh2VD’,! members: [! {name: ‘Alice’, dateJoined: ‘2012-05-24’},! {name: ‘Bob’, dateJoined: ‘2012-05-25’} ! ! !! ]! banned: [! {name: ‘Eve’, dateBanned: ‘2012-05-25’}! !]! }! Alice  and  Bob  are  in  an  awesome  group  
63. groups.update(! {_id: ‘8rOIwh2VD’},! {$push: ! {members: ! {name: ‘Eve’, ! dateJoined: new Date()}! ! },! },! function(err) {! ! !if (err) throw err;! ! !callback();! });! Bugs  in  our  code  fail  to  detect  Eve’s  trickery   and  the  update  query  is  run!  
64. groups: ! {! _id: ‘8rOIwh2VD’,! members: [! {name: ‘Alice’, dateJoined: ‘2012-05-24’},! {name: ‘Bob’, dateJoined: ‘2012-05-25’},! {name: ‘Eve’, dateJoined: ‘2012-05-26’} ! ! !! ]! banned: [! {name: ‘Eve’, dateRequested: ‘2012-05-25’}! !]! }! Blast!    Eve  got  through!    How  can  we  help   prevent  this?  
65. Bake  in  our  assumpSons!   groups.update(! {_id: ‘8rOIwh2VD’, ! ‘members.name’: {$ne: ‘Eve’},! ‘banned.name’: {$ne: ‘Eve’}},! {$push: ! {members: ! {name: ‘Eve’, ! dateJoined: new Date()}! ! },! },! function(err, updateCount) {! if (err) throw err;! if (updateCount !== 1) throw new Error(‘bugs!’);! callback(updateCount === 1);! });!
66. Always  Specify  Fields  
67. To  assign  a  player  a  new  chain,  we  only  need   their  _id  and  game  informaSon   players   account   info   login   chainHistory   stats   game   game  
68. To  load  a  player’s  profile  content,  we  just   need  their  history,  stats,  and  profile  info   players   account   info   login   chainHistory   stats   game   chainHistory   stats   info  
69. We  need  to  assign  Bob  a  new  chain,  lets   figure  out  what  state  he  needs  next   db.players.find({name: ‘Bob’});! ! This  will  fetch  and  send  back  the  ENTIRE   player  object!  
71. Lets  specify  that  we  only  want  our  “game”   subdocument   db.players.find({name: ‘Bob’}, {fields: [‘game’]});! ! Hooray!    Much  less  to  retrieve  and  send   over  the  series  of  tubes.  
72. Store  Everything!  
73. If  you  have  informaSon  available,  save  it   even  if  you  don’t  plan  on  using  it!   db.chains.update({_id: ‘2VmVO18hs’}, ! {$push: {player_id: ‘1qyY61Lu1’,! state: ‘draw’,! date: new Date(),! content: {time: 106896,! ! ! ! ! count: 27,! ! ! ! ! width: 520,! ! ! ! ! height: 390,! ! ! ! ! step_id: ‘1i7RlFbgU’! }! });!
74. Some  Pain  Points  
75. Less  obvious  database  structure   mysql> DESC players;! +----------------+---------+------+-----+---------+-------+! | Field | Type | Null | Key | Default | Extra |! +----------------+---------+------+-----+---------+-------+! | id | int(11) | NO | PRI | NULL | |! | activeStateId | int(11) | YES | | NULL | |! | activeStepId | int(11) | YES | | NULL | |! +----------------+---------+------+-----+---------+-------+!
76. Data  integrity  up  to  applica]on   Mysql> ALTER TABLE players ! ADD CONSTRAINT fk_players_chains! FOREIGN KEY (activeChainId) REFERENCES chains(id); ! /* Oh no! This is going to mess things up! */! ! players.update({_id: ‘c58D4’}, ! {$set: {‘game.activeChain_id’: ‘bad_id’}});! !
77. No  mature  GUI  query  development  tools  
78. If  we  could  start  over  would  we  s]ll  use  MongoDB?  
79. Absolutely!   •  NoSQL  in  general  is  great  for  rapid  prototyping   •  FantasSc  performance     •  IntuiSve  language  keeps  it  easy  to  run  one-­‐off   queries  /  refactor  schema     •  Excellent  (and  now  officially  supported)  Node   driver  
80. Ques]ons?   @DoodleOrDie   @Zugwalt