MongoDB Paris 2012: MongoDB on Azure

Pierre Couzy, Architect, Microsoft
Running MongoDB on a cloud platform has unique constraints, and we've been through these during the last 8 months on a large-scale MongoDB farm. Deployment, monitoring, gotchas, tips & tricks: 100% real-life mongo, questions more than welcome !

Transcript

1. MongoDB'on'Azure' Pierre'Couzy,'Microso4'France'
2. Le'contexte' 2010$ 20'Employees' €1,5m'revenue' 1'game'in'running'mode' 1'game'in'development' Installa+ons:$50M$ MAU:$5M$ DAU:$1M$ 2012$ 80'Employees' €5m'revenue'FYTD' 2'games'in'running'mode' 4'games'in'development' 3'offices:'Paris,'Madrid,'Berlin'
3. Croissance'iniLale'du'jeu'
4. Les'jeux'chez'Kobojo' 2009':'Hébergement'classique'low'cost'(Goobox)' –  Difficulté'dans'la'gesLon'des'pics' 2010Y11':' –  SorLe'de'PyramidVille' •  Croissance'très'forte,'très'rapide' –  IAAS'sur'Amazon'(PyramidVille)' 2012':' –  PAAS'sur'Azure'(AtlanLs)' –  IAAS'sur'Azure'
5. Architecture'cloud'classique' ' ' Service' LB$ Internet' Web'Site' (ASPX,'ASMX,'WCF)' Web'Site' (ASPX,'ASMX,'WCF)' Appli'Web' Worker' Service' Backend' LB$ Data' Stockage' Blobs' Navigateur,' Tablece,' Appli,'…'
6. Architecture'simplifiée' .NET'+'Flash' NOSQL'' SQL'Server' LB' Supervision'
7. Principales'contraintes' mode'KV'Y>'mode'document' '' ' ' 'monolithique'Y>'shardé' ' 'RaLo'écriture/lecture'>'1' ' ' 'Montée'en'charge'rapide'
8. Découper'la'charge' •  Main' Game'data' •  VolaLle' Social'data' •  Locks…'
9. Disques'persistents'Azure'
10. Sharding'
11. Gérer'le'load'balancer'd’Azure'
12. Le'coût'des'replicasets' Routing(And(Config(Role Instance(with(Azure( Drive Instance ReplicaSetRole1_IN_0 ReplicaSetRole1_IN_1 ReplicaSetRole1_IN_2 ReplicaSetRole2_IN_0 ReplicaSetRole2_IN_1 ReplicaSetRole2_IN_2 RoutingConfigRole_IN_1 RoutingConfigRole_IN_0 RoutingConfigRole_IN_2
13. Sharding'sans'replicaset' Routing(And(Config(Role Instance(with(Azure( Drive RoutingConfigRole_IN_1 RoutingConfigRole_IN_0 RoutingConfigRole_IN_2 Shard002 Server(A Server(B Server(C Server(D Shard001 Azure(Table Intance(Ip,(ShardName,(DrivePath Ip(A),(Shard001,(Shard001.vhd Ip(B),(Shard002,(Shard002.vhd
14. Monitoring'
15. Monitoring'' •  ULlisaLon'du'Service'Bus'Azure' •  Déport'Offsite'des'logs' •  ExploitaLon'dans'Graylog2'
16. Fine'tuning' •  GesLon'des'chunks' •  GesLon'des'tailles'de'machine' •  GesLon'des'Lmeouts' •  PoliLque'de'retry' •  Taille'des'pools' •  Mécanismes'd’alerte'automaLque'
17. Les'machines'Azure' Instance' Size' CPU'Cores' CPU' Speed' Memory' Instance' Storage' I/O' Performan ce' Cost/Hour' Extra' Small' Shared' 1.0'GHz' 768'MB' 20'GB' Low' $0.02' Small' 1' 1.6'GHz' 1.75'GB' 225'GB' Moderate' $0.12' Medium' 2' 1.6'GHz' 3.5'GB' 490'GB' High' $0.24' Large' 4' 1.6'GHz' 7'GB' 1,000'GB' High' $0.48' Extra' Large' 8' 1.6'GHz' 14'GB' 2,040'GB' High' $0.96'
18. En'conclusion' •  Commencez'peLt' Le'plus'simple':'azure'IaaS' hcp://www.mongodb.org/display/DOCS/MongoDB +on+Azure' Custom' •  Instrumentez'tôt' mms'a'minima' •  Apprenez'