Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

AWS Cloud School - London April 2012

385 vues

Publié le

Publié dans : Technologie, Business
  • Login to see the comments

  • Soyez le premier à aimer ceci

AWS Cloud School - London April 2012

  1. 1. AWS CLOUD SCHOOL London - Thursday 12th April, 2012
  2. 2. Hello.
  3. 3. Thank you.
  4. 4. StorageTools & ComputeSupport Databases
  5. 5. Consumer Seller business business
  6. 6. Decades of experience Operations, management and scale
  7. 7. Programmatic access
  8. 8. Unexpected innovation
  9. 9. Blinding flash of the obvious
  10. 10. 6 years youngAmazon S3 launched on March 14th, 2006
  11. 11. Objects in S3Billions of objects 905 B 1000.000 750.000 500.000 250.000 0 Q4 2006 Q4 2007 Q4 2008 Q4 2009 Q4 2010 Q4 2011 Q1 2012 650k+ peak transactions per second
  12. 12. Utility computing Available. On-demand. Flexible.
  13. 13. Let’s go!
  14. 14. Launch a high availability web appReliable. Scalable. On-demand. From anywhere. 10 minutes.
  15. 15. cloudbyexample.net
  16. 16. What just happened?
  17. 17. AP LICAT P ION ARCH ECTU IT RE
  18. 18. AP LICAT P ION ARCH ECTU IT RE Load balancer Application servers Data store
  19. 19. AP LICAT P ION ARCH ECTU IT RE Load balancer Application servers Data store Amazon DynamoDB
  20. 20. AP LICAT P ION ARCH ECTU IT RE Load balancerAmazon EC2 Application servers Data store Amazon DynamoDB
  21. 21. AP LICAT P ION ARCH ECTU IT REElastic Load Balancer Load balancerAmazon EC2 Application servers Data store Amazon DynamoDB
  22. 22. AP LICAT P ION ARCH ECTU IT RE Load balancer Application servers Data store
  23. 23. Cloud architecture best practice
  24. 24. “Everything fails, all the time”
  25. 25. Design for failure
  26. 26. AP LICAT P ION ARCH ECTU IT RE Load balancerDecoupled Application servers Data store
  27. 27. AP LICAT P ION ARCH ECTU IT RE Load balancer Stateless Application servers Data store
  28. 28. AP LICAT P ION ARCH ECTU IT RE Load balancerRedundancy Application servers Data store
  29. 29. AP LICAT P ION ARCH ECTU IT RE Load balancerRedundancy Application servers Data store
  30. 30. AP LICAT P ION ARCH ECTU IT RE Load balancerAvailability zone Application servers redundancy Data store
  31. 31. AP LICAT P ION ARCH ECTU IT RE Load balancerHorizontal scale Application servers Data store
  32. 32. AP LICAT P ION ARCH ECTU IT RE Load balancerHorizontal scale Application servers Data store
  33. 33. AP LICAT P ION ARCH ECTU IT RE Load balancerHorizontal scale Application servers Data store
  34. 34. Design for elasticity Scale out. Scale in.
  35. 35. Usage Midnight 7am 9am Noon 3pm 5pm Midnight Time of day
  36. 36. Usage January December Time of year
  37. 37. Usage January December Time of year
  38. 38. Usage 60% January December Time of year
  39. 39. Design for cost Pricing options.
  40. 40. Economies of scale
  41. 41. 19 price dropsCommitted to passing savings to customers
  42. 42. Utilisation
  43. 43. Achieving economies of scale100% Time
  44. 44. Achieving economies of scale100% Reserved capacity
  45. 45. Achieving economies of scale100% On-demand Reserved capacity
  46. 46. Achieving economies of scale100% On-demand Reserved capacity
  47. 47. Spot marketChoose your own price for compute
  48. 48. Fine grainedtime/cost benefit
  49. 49. 30,472 cores $1,279 per hour
  50. 50. AP LICAT P ION ARCH ECTU IT RE Load balancer Application servers Data store
  51. 51. AP LICAT P ION ARCH ECTU IT RE Load balancer Application serversCloudFormation template Data store
  52. 52. AP LICAT P ION ARCH ECTU IT RE Load balancer Application serversCloudFormation template Data store
  53. 53. AWS CloudFormationAutomated provisioning tool. Template based.
  54. 54. Atomic and idempotent Also: free!
  55. 55. CloudFormation templateCloudFormation service
  56. 56. CloudFormation template CloudFormation serviceEC2 instances Load balancer
  57. 57. EC2 instance sizesRange of compute, storage and memory Metered by Elastic Compute Unit
  58. 58. Standard High memory High CPU m1 m2 c1
  59. 59. 1 instance for 100 hours =100 instances for 1 hour You choose where to balance cost against time.
  60. 60. Placement groupsCluster compute10 gig E, non blocking network
  61. 61. 42nd240 TFLOPS. World’s fastest ethernet supercomputer.
  62. 62. GPU 2 x NVIDIA Tesla. 400 cores each.Molecular modeling. Hedging. Transcoding and rendering.
  63. 63. CloudFormation template CloudFormation serviceEC2 instances Load balancer
  64. 64. CloudFormation template CloudFormation serviceEC2 instances Load balancer S3 bucket
  65. 65. 99.999999999% durability
  66. 66. CloudFormation template CloudFormation serviceEC2 instances Load balancer S3 bucket
  67. 67. CloudFormation template CloudFormation serviceEC2 instances Load balancer S3 bucket Users and groups
  68. 68. CloudFormation template CloudFormation serviceEC2 instances Load balancer S3 bucket Users and Autoscaling groups
  69. 69. CloudFormation template CloudFormation serviceEC2 instances Load balancer S3 bucket Users and Autoscaling groupsSecurity groups Port access Write access Scaling parameters
  70. 70. Identity and Access Management
  71. 71. API level restrictions Restricted by default
  72. 72. Account
  73. 73. BillingAccount credentials Account MFA
  74. 74. AccountDBA Developer Sys admin Finance Roles
  75. 75. AccountDBA Developer Sys admin Finance Roles Sally Robert Users Chris
  76. 76. Security credentials Multifactor authenticationManagement console access Data read/write access API level access
  77. 77. AccountDBA Developer Sys admin Finance Roles Sally Robert Users Chris
  78. 78. Restricted credentialsAccess specific S3 bucket, DynamoDB and CloudFormation
  79. 79. A note on security...
  80. 80. Shared responsibility
  81. 81. Service Organisation Controls 1 ISO 27001Shared responsibility FISMA PCI DSS Moderate Level 1
  82. 82. Data stays local
  83. 83. aws.amazon.com/security
  84. 84. Autoscaling
  85. 85. Automatic scalingAvailability zone and load balancer aware.
  86. 86. MetricsCPU, network, IO, custom
  87. 87. Metrics CPU, network, IO, custom CloudWatch alarmsSet operational thresholds: IOPS, page load time
  88. 88. Metrics CPU, network, IO, custom CloudWatch alarms Set operational thresholds: IOPS, page load timeTrigger autoscaling policy Adjust autoscaling groups within bounds Register with load balancer
  89. 89. AP LICAT P ION ARCH ECTU IT RE Load balancer Application servers Data store Amazon DynamoDB
  90. 90. Any database on Amazon EC2 MySQL, DB2, Oracle, PostgreSQL...
  91. 91. Relational Database Service Managed MySQL and Oracle databases
  92. 92. Rapid High provisioning availabilityScalable Scalablestorage compute Relational Database Service Managed MySQL and Oracle databases
  93. 93. High performance databases Increase throughput Increase availability Reduce latency
  94. 94. High performance databases Read replicasPush-button scaling Increase throughput ElastiCache Increase availability Reduce latency
  95. 95. High performance databases Increase throughputMulti-AZ Increase availability Reduce latency
  96. 96. High performance databases Increase throughput Increase availability Reduce latency ElastiCache
  97. 97. ProblemPerformance decreases at scale
  98. 98. Performance Predictable, consistent Scale
  99. 99. Performance Predictable, consistent Degraded performance with scale Scale
  100. 100. Performance Predictable, consistent Degraded performance with scale Scale
  101. 101. = more problems
  102. 102. Data caching Provisioning!Data sharding = more problemsCluster management Fault management
  103. 103. Undifferentiated heavy lifting
  104. 104. DynamoDB
  105. 105. Fully managedNoSQL database service
  106. 106. Extremely fastperformance
  107. 107. Consistently fast performance
  108. 108. Low latency Single digit millisecond
  109. 109. SSD backed Low latency Single digit millisecond< 5 ms reads < 10 ms writes
  110. 110. Seamless scalability No table size limits. Unlimited storage. Live repartitioning. Zero admin.
  111. 111. Durable and available Consistent, disk-only writes
  112. 112. Predictable performance Provisioned throughput
  113. 113. Reserve required IOPS Per table. Set at creation. Scale via API.
  114. 114. Scale at any time No downtime
  115. 115. Data model Flexible. Schema-less.
  116. 116. Simple key/value pairs title => “Introduction to DynamoDB” date => “20120320”
  117. 117. Associative array, or Hash[ title => “Introduction to DynamoDB”, date => “20120320” ]
  118. 118. Attributes[ title => “Introduction to DynamoDB”, date => “20120320” ]
  119. 119. [ title => “Disaster Recovery with AWS”, date => “20120320”, format => “online seminar”, presenter => “Jeff Barr” ] Attributes [ title => “Introduction to DynamoDB”, date => “20120320” ]
  120. 120. [ title => “Disaster Recovery with AWS”, date => “20120320”, format => “webinar”, presenter => “Jeff Barr” ] Items [ title => “Introduction to DynamoDB”, date => “20120320” ]
  121. 121. [ title => “Disaster Recovery with AWS”, date => “20120328”, format => “webinar”, presenter => “Jeff Barr” ] Table [ title => “Introduction to DynamoDB”, date => “20120320” ]
  122. 122. Table
  123. 123. Item “UserID” = “1” “Date” = “20100915”“Title” = “flower”“Tags” = “flower”,“jasmine”, “white”
  124. 124. “UserID” = “1” “UserID” =”2” “UserID” =”3” “Date” = “Date” = “Date” = “20100915” “20100916” “20100917”“Title” = “flower” “Title” = “ferrari” “Title” = “coffee”“Tags” = “flower”, “Tags” = “car”, “Tags” = “drink”,“jasmine”, “white” “italian” “delicious”
  125. 125. “ImageID” = “1” Primary or hash key“Date” = “20100915” “Title” = “flower” “Tags” = “flower”, “jasmine”, “white”
  126. 126. “ImageID” = “1” Primary or hash key“Date” = “20100915” Composite or range key “Title” = “flower” “Tags” = “flower”, “jasmine”, “white”
  127. 127. “ImageID” = “1” Primary or hash key“Date” = “20100915” Composite or range key “Title” = “flower” “Tags” = “flower”, Sets of strings “jasmine”, “white” or numbers
  128. 128. Best practice Well balanced, fine grained hash keys.Customer, order, item, etc. rather than store_id.
  129. 129. Simple API Only 12 operations.
  130. 130. Consistency Writes are always consistent.Implicit, item-level transactions.
  131. 131. Durability Writes occur to disk, not memory.Writes are acknowledged once they have been made in two physical data centres.
  132. 132. Availability Region specific (not AZ)Continuously replicated across multiple AZs
  133. 133. Zero adminSingle API call deployment and scaling.
  134. 134. AP LICAT P ION ARCH ECTU IT RE Load balancer Application servers Data store
  135. 135. High availabilityDecouple. Horizontal scale. Availability zones.
  136. 136. AutomateRapid, error free provisioning.
  137. 137. DynamoDBNo administration NoSQL.
  138. 138. What’s next?
  139. 139. Analytics with Elastic MapReduce Built for data. Designed for humans.
  140. 140. create external table items_db (id string, likes bigint, views bigint)stored byorg.apache.hadoop.hive.dynamodb.DynamoDBStorageHandlertblproperties ("dynamodb.table.name" = "items", "dynamodb.column.mapping" = "id:id,likes:likes,views:views");select id, likes, views from items_dborder by views desc;
  141. 141. Simple Workflow Managed workflows
  142. 142. Create thumbnail Update Top 100New share Upload image Duplicate detection
  143. 143. Amazon CloudSearchManaged search. Provide documents. Search. Up and running in an hour.
  144. 144. Create search domain Upload documentsSearch Document Format (SDF) XML or JSON Search!
  145. 145. Sophisticatedapplications
  146. 146. Reliableapplications
  147. 147. Innovativeapplications
  148. 148. Undifferentiated heavy lifting
  149. 149. Productivity
  150. 150. Left image Right image Warp Warp Stereo match 3D map Tiling
  151. 151. Thank you!
  152. 152. Q&Amatthew@amazon.com
  153. 153. TutorialsSelf-directed. Choose your track.
  154. 154. Hands-on sessions
  155. 155. Break out sessions
  156. 156. AWS staff
  157. 157. cloudbyexample.net

×