The document discusses the integration of NoSQL, specifically Apache Cassandra, into Constant Contact's operations to handle increasing data needs from small businesses for social media marketing. It addresses the challenges and solutions related to implementation, operational efficiencies using Puppet for DevOps, and the benefits of Cassandra's fault-tolerant and elastic data model. Overall, it highlights the evolution from traditional infrastructure to a more agile and cost-effective system, enabling rapid deployment and monitoring.

1. Cassandra & Puppet:Scaling data at $15/monthConstant ContactMarch 2011Dave Connors – VP OperationsJim Ancona – Systems ArchitectMark Schena – Manager Systems Automation

2. Constant ContactConstant Contact2000 – 2010 Market leader for Small BusinessesEmail, Event & Survey

3. Over 400k paying customers

4. No. 134 on the Deloitte Technology Fast 500 listingBusiness modelMany customers pay as little as $15 a month

5. ~2 million database transactions per minuteConstant ContactThe business problem

6. Constant Contact Small Businesses are looking to us for help with Social Media marketingSocial Media 10-100 times more data

7. Challenge with our business modelThe Key ChallengeThe Key ChallengeIntegrate social media dataSolution = NoSQL

8. Cost = Low

9. Time to market = ?ImplementationImplementing NoSQLOps and Dev both face issuesData model

10. Monitoring

11. Authentication

12. Logging

13. Risk profile

14. Roles & ResponsibilitiesOpsDev

15. Apache CassandraApache CassandraDeveloped at Facebook

16. Open sourced in 2008

17. Incubated at Apache

18. Became an Apache top-level project in 2010

19. http://cassandra.apache.org

20. In use at Digg, Facebook, Twitter, Reddit, Rackspace, Cloudkick, Cisco, …

21. Largest production cluster has over 100 TB of data in over 150 machinesWhat is Cassandra?What is CassandraImplemented in Java

22. Fault Tolerant

23. Elastic

24. Durable

25. Rich data model

26. Replicated data

27. Consistency optionsReplicationReplicationHow many copies of each piece of data do we want?N=3

28. Consistency LevelONEConsistency Level OneY

29. Consistency Level QuorumX

30. Risks and MitigationRisks and MitigationMoving target

31. Developer unfamiliarity

32. Operational procedures

33. Reliability concerns

34. Deployment automation

35. Community involvement

36. Training/Consulting

37. Application selection

38. Lots of monitoring

39. Phased rolloutDevelopment ChallengesDevelopment ChallengesUnderstanding the data modelChoosing a clientClients available for Java, Python, .NET, Ruby, PHPDon’t use ThriftMoving target

40. Open SourceNot “one neck to wring”

41. Paid support and training is available: http://datastax.com

42. CommunityMailing listsIRC #cassandra at freenodeContributePhased RolloutSwitchable modes

43. Mirroring

44. Dial-able traffic CollaborationBig, complex project

45. Close collaboration

46. Flexible roles

47. Ability to iterateOpsDev

48. “Are you sure you really want that?” “Are you sure you really want that?”3 500G disks

49. 1 250G disk

50. No SWAP

51. RAID Zero Root Partition and Data Storage

52. 32G MemoryWe will need how many servers?We will need how many servers?

53. How many nodes?Quorum = 3

54. Multiple Datacenters = 2

55. Use only half the available disk = 2

56. 12 Servers = ~1 TB Of Data Storage

57. ~6 TB of Data Storage 72x 6 =x 2 = 123x 2 = 6

58. RanRandom Partitioner

59. Tool ChainTool Chain

60. with PuppetPuppet is the shared framework between Operations and Development

61. Versioning of puppet code allows for adoption of development best practices

62. Leverage Domain specific knowledge and skillDevOps with Puppet

63. Always Move ForwardAlways Move Forward

64. Operational EfficienciesRemote logging is a requirement

65. Cassandra uses log4j natively

66. Resources not available for remote log4j development

67. Scribed with Puppet provides the solutionOperational Efficiencies

68. Development takes the Operational LeadMunin

69. JMX trending

70. Identify critical data points

71. Rapid development of graphs

72. Puppet Definitions are used for rapid deploymentSample Munin Graph

73. Puppet Code Example: Munin Puppet Codedefine munin::cassandracolumnfamily ( ) { include cassandravirtual File <| title == "jmxbin" |> $confdir="/opt/cassandra-munin-plugins” $plugindir="/etc/munin/plugins" $target="/opt/cassandra-munin-plugins/jmx_" # Match 3 strings separated by periods $pattern = '^([^.]*)[.]([^.]*)[.]([^.]*)$' $keyspace = regsubst($name, $pattern, '\1') $columnfamily = regsubst($name, $pattern, '\2') $file = regsubst($name, $pattern, '\3')file {"${keyspace}_${columnfamily}_${file}.conf": owner => 'root', ensure => 'file', group => 'root', type => 'file', path => "${confdir}/${keyspace}_${columnfamily}_${file}.conf", mode => '644', content => template("munin/attribute_${file}.conf.erb"), require => [ Package['munin-node'], File['/opt/cassandra-munin-plugins'], File['jmxquery'], ], }file {"$plugindir/${keyspace}_${columnfamily}_${file}": ensure => 'link', owner => 'root', group => 'root', mode => '511', type => 'link', target => "$target", require => [ File['/opt/cassandra-munin-plugins'], File["${keyspace}_${columnfamily}_${file}.conf"], File['jmxquery'], Package['munin-node'], ],

74. ConclusionConclusionCassandra as an appliance

75. Development Best Practices with Life Cycle Management

76. Traditional vs. Today

77. Infrastructure 4 weeks 4 hours to build 72 nodesDevelopment to Deployment 9 months 3 monthsCost Millions 150k

78. Q&AThank You!