innovations_and_trends_in_cloud.pdf

Transcript

1. © 2015, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Julio M. Faerman @jmfaerman AWS Meetup Mexico Innovations and Trends in Cloud Computing
2. How AWS works under the hood?
3. What is new and trending?
4. Tenets of AWS Capacity
5. On demmand
6. Reliability On demmand
7. Reliability Low cost On demmand
8. Scale
9. Not that Long Ago…
10. Not that Long Ago… • 11-15% utilization
11. Not that Long Ago… • 19% utilization
12. Not that Long Ago… • 19% utilization • 81% idle
13. Not that Long Ago… • 19% utilization • 81% idle • Lots of meetings
14. Not that Long Ago… • 19% utilization • 81% idle • Lots of meetings • 3 to 9 months to deploy apps
15. Partial reliability
16. Partial reliability Unnecessary and missing features
17. Partial reliability Unnecessary and missing features Outages
18. Poor Utilization
19. Capacity not Available
20. Vicious Cycle
21. What changes with scale?
22. Land and Power
23. Power
24. Power Infrastructure • Power generation facilities • Negotiated PPAs • Commitment to renewable energy sources
25. Amazon Solar Farm US East • East Virginia • June 2015 • Community Energy • 80 MW solar
26. Amazon Wind Farm US East • North Carolina • July 2015 • 208 MW wind
27. Amazon Wind Farm (Fowler Ridge) • Indiana • January 2015 • Pattern Development • 150 MW wind
29. November 2014 Announce commitment to 100% renewable energy April 2015 Achieved 25% renewable milestone Announced 4.8 MWh battery pilot with Tesla April 2015 Joined the American Council on Renewable Energy (ACORE) April 2015 Announced participation in the U.S. Partnership for Renewable Energy Finance End of 2016 Goal of 40% renewable energy for power consumed AWS Renewable Energy Commitment
30. Power at Scale
31. How is that more efficient?
32. Cloud Efficiency Average AWS Customer… • Uses 77% fewer servers • Uses 84% less power
33. Cloud Efficiency NRDC 2014 assessment… • Cloud companies at 65% utilization
34. Cloud Efficiency NRDC 2014 assessment… • Cloud companies at 65% utilization • On premises at ~15% utilization
35. Cloud Efficiency NRDC 2014 assessment… • Cloud companies at 65% utilization • On premises at ~15% utilization • 15 ÷ 65 = 23%
36. Cloud Efficiency Annual Uptime Institute survey…
37. Cloud Efficiency Annual Uptime Institute survey… • PUE of 1.7 = enterprise datacenter
38. Cloud Efficiency Annual Uptime Institute survey… • PUE of 1.7 = enterprise datacenter • PUE of 1.07 to 1.15 = cloud datacenter
39. Cloud Efficiency Annual Uptime Institute survey… • PUE of 1.7 = enterprise datacenter • PUE of 1.07 to 1.15 = cloud datacenter • 1.2 ÷ 1.7 = 71% • A 29% reduction in power pulled from the grid
40. Cloud Efficiency What this means… • 23% x 71% = 16% • 84% reduction in power
41. • ~ 50% renewable • Server utilization >50% • Reduced server costs 85% • 115+ million members • 33%~ US Downstream • 42.5B hours streamed in 2015 AWS re:Invent 2015 | A Day in the Life of a Netflix Engineer ”10s of operation engineers”
42. Planning, Design & Construction
43. Security
44. Security
45. Security
46. Security
47. Security
48. Security
49. Redes
50. Old World Why it doesn’t work… • Complex hardware / software • Burdened with unnecessary features • Buggy, slow to fix, slow to improve • Complexity equals cost
51. Need for a Purpose Build Network Advantages of purpose-built networks: • Simple designs at each layer • Hardware & software not burdened with unnecessary features • Simple software = less hardware • Simplicity = lower cost
52. AZ AZ AZ AZ AZ Purpose Built Network • Each datacenter has a purpose built network
53. AZ AZ AZ AZ AZ Purpose Built Network
54. AZ AZ AZ AZ AZ Purpose Built Network
55. AZ AZ AZ AZ AZ Purpose Built Network • Metro-area DWDM links between AZs • 82,864 fiber strands in region • AZs < 2ms apart & usually <1ms • 25+ Tbps inter-AZs peak capacity • Each datacenter has a purpose built network
56. AZ AZ AZ AZ AZ Transit Transit • Redundant paths to transit centers • Transit Centers connect to: • Private links to other AWS regions • Private links to AWS Direct Connect customers • Internet through peering & paid transit Purpose Build Network • Metro-area DWDM links between AZs • 82,864 fiber strands in region • AZs <2ms apart & usually <1ms • 25+ Tbps inter-AZs peak capacity • Each datacenter has a purpose built network
57. AWS Worldwide Network Backbone • Provides full, direct control • Reduces cost, complexity, jitter • Increases performance & availability • Better Quality of Service (QoS) • Not over-subscribed and fully alarmed
58. Servers
59. Need for Customization Advantages of purpose-built servers: • Work directly with component manufacturers like memory, HDD, CPU • Design optimized for specific workloads • Hardware & software not burdened with unnecessary features • Highly utilized
60. Custom High Density Storage Rack • 864* disks • 1065* Kg
61. Importance of Supply Chain
62. Importance of Supply Chain
63. Tailand floods in 2011
64. Tailand floods in 2011
65. Importance of Supply Chain
66. Importance of Supply Chain
67. Importance of Supply Chain
68. Retail Leverage
69. AWS Worldwide Locations India Ohio Ningxia Montreal UK
70. * As of 1 Mar 2016 2009 48 280 722 82 2011 2013 2015 Ritmo de Inovação Acelerado
71. http://blog.concretesolutions.com.br/2015/04/amazon-machine-learning-benchmark/ http://blog.concretesolutions.com.br/2015/04/amazon-machine-learning-benchmark/
75. AWS IoT DEVICE SDK Set of client libraries to connect, authenticate and exchange messages DEVICE GATEWAY Communicate with devices via MQTT and HTTP AUTHENTICATION AUTHORIZATION Secure with mutual authentication and encryption RULES ENGINE Transform messages based on rules and route to AWS Services AWS Services - - - - - 3P Services DEVICE SHADOW Persistent thing state during intermittent connections APPLICATIONS AWS IoT API DEVICE REGISTRY Identity and Management of your things
76. AWS Lambda You can now develop your AWS Lambda function code using Python, maintain multiple versions of your function code, invoke your code on a regular schedule, and run your functions for up to five minutes. AWS Database Migration Service Migrate databases to AWS with minimal downtime AWS Database Migration Service helps you migrate databases to AWS easily and securely with minimal downtime. It supports all widely used database platforms, and performs schema and code conversion for migrations between database engines. AWS Database Migration Service is low cost and simple to use. You only pay for the compute resources used during the migration process and can typically migrate a terabyte-size database for as little as three dollars. Amazon Kinesis Firehose Easily load streaming data into S3 & Redshift Amazon Kinesis Firehose is the easiest way to load streaming data into AWS. It can capture and automatically load streaming data into Amazon S3 and Amazon Redshift, enabling near real- time analytics with existing business intelligence tools you’re already using today. It is a fully managed service that requires no ongoing administration.
77. Em resumo • Escala e ritmo de inovação sem precedentes • Demanda personalização e inovação em todas dimensões de um datacenter • O valor entregue aos cliente se extende além do hardware e software
80. Thank You Gracias! Julio M. Faerman @jmfaerman