exploring a new way to manage systems with ostree and
play

Exploring a new way to manage systems with ostree and atomic - PowerPoint PPT Presentation

Exploring a new way to manage systems with ostree and atomic Michael Scherer, misc@redhat.com Who am I ? Sysadmin @ In a previous episode.. Packager/sysadmin/build system maintainer @mageia Python coder @ some firewall startup Counter


  1. Exploring a new way to manage systems with ostree and atomic Michael Scherer, misc@redhat.com

  2. Who am I ?

  3. Sysadmin @

  4. In a previous episode..

  5. Packager/sysadmin/build system maintainer @mageia

  6. Python coder @ some firewall startup

  7. Counter stri^W Sysadmin @ unamed european institution

  8. Old ways

  9. Assemble set of packages

  10. Binary packages

  11. Sources

  12. Both ways

  13. Base + ports

  14. Appliances..

  15. .. or phones/tablets

  16. A new hop^W way

  17. Project Atomic

  18. Codename for a set of distribution

  19. Centos/Fedora

  20. Others are welcome

  21. Atomic upgrade

  22. Containers

  23. Lightweight

  24. Base of openshift v3

  25. Let's get technical

  26. Components, bottom To top

  27. Ostree

  28. Created for testing GNOME

  29. Firmware-like

  30. Bootable read only filesystem

  31. All in /usr + some link to /var

  32. Reboot on new version

  33. Reboot on old version

  34. Make rollback possible

  35. Git for filesystem

  36. Docker & systemd

  37. Ship applications with runtime

  38. API to start, monitor, control services

  39. Cgroups usage to limit containers

  40. Geard / Kubernetes

  41. Orchestrate containers

  42. Manage network

  43. Distribute tasks

  44. Ensure redundancy

  45. Various others things

  46. SELinux

  47. Isolation

  48. Protect base system from containers

  49. Protect containers from containers

  50. Cherry on the cake

  51. Cockpit as a web interface

  52. Why ?

  53. The stars aligned right

  54. Ostree from Gnome QA

  55. Jails => Openvz => Linux namespace => LXC => Docker

  56. SELinux / svirt

  57. Systemd

  58. Pressure on IT departments

  59. Automate more

  60. Drive human costs down

  61. Colocate services more

  62. Drive hardware spending down

  63. Deliver faster

  64. Drive profits high

  65. Agility

  66. Dev Ops

  67. Dev systems != Prod servers

  68. Example

  69. Gluster.org web infra

  70. Need to be redone

  71. Static website

  72. Need a httpd server

  73. Need a builder

  74. Need a wiki

  75. Mediawiki

  76. MySQL || PostgreSQL

  77. Varnish

  78. Awstats

  79. Traditional way

  80. All on 1 server

  81. VM way

  82. 1 VM per service

  83. Containers way

  84. Everything in a different containers

  85. Microservices

  86. Why ?

  87. A few problems

  88. Static website build

  89. Use middleman

  90. Ruby

  91. No comment

  92. Web designer use F20

  93. So better use this as well

  94. Unpackaged software

  95. Mediawiki

  96. Need specific isolation

  97. Test upgrade and rollback

  98. Scaling issues

  99. Mediawiki write on the FS

  100. Use S3 backed storage

Recommend


More recommend