Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

High availability in IT: AAAARGH

4,185 views

Published on

My Barcamp presentation (#bca4) on High Availability, and how much energy and frustration it costs to implement it. In Dutch.

Published in: Technology
  • Be the first to comment

High availability in IT: AAAARGH

  1. 1. High Availability in IT Mattias Geniar BarCamp 2010 High Attention High Annoyance High AAAAAARGH
  2. 2. High Availability <ul><li>Introduction
  3. 3. Name: Mattias Geniar
  4. 4. Twitter: @mattiasgeniar
  5. 5. Blog: http://mattiasgeniar.be
  6. 6. E-mail: m@ttias.be
  7. 7. Job: System Engineer @ Nucleus </li></ul>
  8. 8. Walkthrough <ul><li>What is High Availability?
  9. 9. What is “downtime”
  10. 10. How are systems currently built </li><ul><li>What are the weak points
  11. 11. How can we prevent it </li></ul><li>What's so annoying about it </li></ul>
  12. 12. #define ha <ul><li>What is high availability? </li></ul>
  13. 13. #define ha <ul><li>What is High Availability? </li></ul><ul>High availability is a system design approach and associated service implementation that ensures a prearranged level of operational performance will be met during a contractual measurement period . </ul>
  14. 14. #define ha <ul><li>What is High Availability? </li></ul><ul>We can use Facebook, Twitter, Work, Blog, ... </ul>
  15. 15. #define ha <ul><li>What is High Availability? </li></ul><ul>Redundancy All good things come in pairs. </ul>
  16. 16. #define ha <ul><li>What is High Availability? </li></ul><ul>Backup A working backup, when you need it. </ul>
  17. 17. #define ha <ul><li>What is High Availability? </li></ul><ul>Connectivity You can reach your system, when you need it. </ul>
  18. 18. #define downtime <ul><li>When is it considered downtime? </li></ul>
  19. 19. #define downtime <ul><li>When is it considered downtime? </li></ul><ul>When it's unavailable site offline, application unreachable, ... </ul>
  20. 20. #define downtime <ul><li>When is it considered downtime? </li></ul><ul>When it's partially unavailable site is working, but some key functions are malfunctioning </ul>
  21. 21. #define downtime <ul><li>When is it considered downtime? </li></ul><ul>When it's slow to respond users click away on slow sites, get annoyed at loading times, ... </ul>
  22. 22. #define downtime <ul><li>When is it considered downtime? </li></ul><ul>Planned vs Unplanned Is a planned maintenance “downtime”? </ul>
  23. 23. #define downtime <ul><li>What is acceptable? </li></ul>
  24. 24. #define downtime <ul><li>What is acceptable? </li></ul><ul>Webshop during the night: 30 minutes offline ? </ul>
  25. 25. #define downtime <ul><li>What is acceptable? </li></ul><ul>Business Application during off-hours: 15 minutes offline ? </ul>
  26. 26. #define downtime <ul><li>What is acceptable? </li></ul><ul>Airport during rush hour: 5 minutes offline ? </ul>
  27. 27. #define downtime <ul><li>What is acceptable? </li></ul><ul>Twitter during World Championship: 2 minutes offline ? </ul>
  28. 28. #define downtime <ul><li>What is acceptable? </li></ul><ul>Nuclear Power Plant during scheduled maintenance: 15 seconds offline ? </ul>
  29. 29. #define downtime <ul><li>Impact of downtime? </li></ul>
  30. 30. #define downtime <ul><li>Impact of downtime? </li></ul><ul>Webshop can't sell when they're offline. </ul>
  31. 31. #define downtime <ul><li>Impact of downtime? </li></ul><ul>People ignore you, and move on. </ul>
  32. 32. #define downtime <ul><li>Impact of downtime? </li></ul><ul>Your reputation takes (a) (s)hit. </ul>
  33. 33. Our savior: High Availability <ul><li>No more downtime
  34. 34. “Always on” services
  35. 35. 24/7 availability
  36. 36. Responsive sites
  37. 37. = Increase in productivity = WIN? </li></ul>
  38. 38. Designing for High Availability <ul><li>The datacenter </li></ul>
  39. 39. #define downtime <ul><li>What's a datacenter? </li></ul><ul>A secure area </ul>
  40. 40. #define downtime <ul><li>What's a datacenter? </li></ul><ul>Filled with racks. </ul>
  41. 41. #define downtime <ul><li>What's a datacenter? </li></ul><ul>Who are filled with servers. </ul>
  42. 42. #define downtime <ul><li>What's a datacenter? </li></ul><ul>Which are all kept cool. </ul>
  43. 43. Designing for High Availability <ul><li>The typical datacenter design </li></ul>
  44. 44. High Availability: Datacenter This should be simple. Building POWER COOLING
  45. 45. High Availability: Datacenter UPS Hmm ... Building POWER COOLING POWER COOLING
  46. 46. High Availability: Datacenter UPS ... and done! Building POWER COOLING POWER COOLING UPS GENERATOR GENERATOR UPS
  47. 47. Designing for High Availability <ul><li>More than just power & cooling </li></ul>
  48. 48. High Availability: Datacenter <ul>Cages or seperate rooms. </ul>
  49. 49. High Availability: Datacenter <ul>General security </ul>
  50. 50. Designing for High Availability <ul><li>Your hosting </li></ul>
  51. 51. High Availability: Hosting Right. How hard can this be? INTERNET CONNECTIVITY SERVER
  52. 52. High Availability: Hosting UPS UPS I shouldn't have said that. INTERNET SERVER A NETWORK A INTERNET SERVER C NETWORK B SERVER B
  53. 53. Designing for High Availability <ul><li>Example: very important, high traffic webshop </li></ul>
  54. 54. High Availability: Hosting Go get'm Frank! WEBSITE ORDER
  55. 55. High Availability: Hosting Holy servers, Batman! FRONTEND FRONTEND FRONTEND LOAD BALANCER LOAD BALANCER STORAGE STORAGE DATABASE DATABASE
  56. 56. Designing for High Availability <ul><li>Doesn't that get messy? </li></ul>
  57. 57. High Availability: Hosting <ul><li>Doesn't that get messy? </li></ul><ul>Sometimes. </ul>
  58. 58. High Availability: Hosting <ul><li>Doesn't that get messy? </li></ul><ul>But it doesn't have to. </ul>
  59. 59. Designing for High Availability <ul><li>That's it for hosting then? </li></ul>
  60. 60. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>First, there's backups. Lots of them. </ul>
  61. 61. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>Monitoring – know what's happening. </ul>
  62. 62. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>Monitoring – get alerted. </ul>
  63. 63. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>Monitoring – react. 24/7 </ul>
  64. 64. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>Automate as much as possible. </ul>
  65. 65. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>Think about every move you make. </ul>
  66. 66. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>Innovate. Change. </ul>
  67. 67. High Availability: Hosting <ul><li>That's it for hosting then? </li></ul><ul>Eliminate further (hardware) failures. </ul>
  68. 68. Designing for High Availability <ul><li>The Application Website, Business App, ... </li></ul>
  69. 69. High Availability: Developer <ul>Seperate development environment. </ul><ul><li>The Application </li></ul>
  70. 70. High Availability: Developer <ul>Dedicated & motivated programmers. </ul><ul><li>The Application </li></ul>
  71. 71. High Availability: Developer <ul>Knowledge of underlying systems. </ul><ul><li>The Application </li></ul>
  72. 72. High Availability: Recap <ul><li>Power
  73. 73. Cooling </li></ul><ul><li>Internet connectivity
  74. 74. Servers </li></ul><ul><li>Datacenter </li></ul><ul><li>Hosting Provider </li></ul><ul><li>Application </li></ul><ul><li>Developer </li></ul>We've got it!
  75. 75. High Availability: 99% vs 99,99% <ul><li>99% availability
  76. 76. Monthly downtime: 7,20 hours </li></ul><ul><li>99,99% availability
  77. 77. Monthly downtime: 4,32 minutes </li></ul><ul><li>99,9999% availability
  78. 78. Monthly downtime: 2,59 seconds </li></ul>Numbers?
  79. 79. Presentation Down Mattias Geniar BarCamp 2010 Thanks for not running away.

×