High Availability and Xen

3,518 views

Published on

Björn Brundert Presentation on High Availability

Published in: Technology, Travel
  • Be the first to comment

  • Be the first to like this

High Availability and Xen

  1. 1. Highly available infrastructures based on Xen Björn Brundert
  2. 2. Disclaimer <ul><li>- All statements given in this presentation do not represent the opinion of the speakers company but only the personal opinion and experience of the speaker himself </li></ul>
  3. 3. Introduction, overview and scope <ul><li>Virtualization and high availability </li></ul><ul><li>Hardware requirements and planning </li></ul><ul><li>Available solutions today </li></ul><ul><li>Future prospect </li></ul><ul><li>Summary </li></ul><ul><li>Q&A </li></ul>
  4. 4. 1) Virtualization and high availability <ul><li>Standalone server does not meet HA-SLAs </li></ul><ul><li>HA means redundancy  clustering </li></ul><ul><li>Cluster means higher complexity: building and operating </li></ul><ul><li>Dedicated hardware per HA-service becomes more and more expensive </li></ul><ul><ul><li>Hardware, infrastructure components, ETH/FC ports, rackspace, power, cooling </li></ul></ul><ul><ul><li>Active/passive cluster  unused ressources </li></ul></ul>
  5. 5. <ul><li>How to achieve HA </li></ul>1) Virtualization and high availability Application Operating System Hardware Application Operating System Hardware Xen-Hypervisor Xen-Hypervisor Black Box Black Box 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16
  6. 6. 1) Virtualization and high availability <ul><li>Virtualization offers enourmous benefits </li></ul><ul><ul><li>Consolidation: multiple services on one hardware </li></ul></ul><ul><ul><li>Higher flexibility through VM encapsulation and hardware independence </li></ul></ul><ul><ul><li>Application does not require to be „cluster-ready“ </li></ul></ul>
  7. 7. 2) Hardware requirements and planning <ul><li>Key rules when thinking about HA </li></ul><ul><ul><li>Be as paranoid as you want </li></ul></ul><ul><ul><li>There is never enough redundancy </li></ul></ul><ul><ul><li>Murphy is with you </li></ul></ul><ul><li>Identify single points-of-failure </li></ul><ul><li>Consolidation means also higher impact in case of an outage </li></ul>
  8. 8. <ul><li>Server: </li></ul>2) Hardware requirements and planning
  9. 9. 2) Hardware requirements and planning = ? No!
  10. 10. 2) Hardware requirements and planning <ul><li>- I/O Cards: </li></ul>FC NIC
  11. 11. 2) Hardware requirements and planning <ul><li>Server </li></ul><ul><ul><li>Harddisk: RAID-configuration, spare </li></ul></ul><ul><ul><li>Redundant I/O cards </li></ul></ul><ul><ul><li>Redundant power supplies, redundant power cabeling and redundant power circuits </li></ul></ul><ul><ul><li>Redundant heartbeat-interconnect between nodes </li></ul></ul>
  12. 12. 2) Hardware requirements and planning <ul><li>Network components </li></ul><ul><ul><li>Redundant cabeling to server NICs </li></ul></ul><ul><ul><ul><li>Active/passive bonding needs to be set up in OS </li></ul></ul></ul><ul><li>Redundant switch or redundant switch modules </li></ul>
  13. 13. 2) Hardware requirements and planning <ul><li>Storage components: </li></ul><ul><ul><li>RAID configuration with hot-spare harddrives </li></ul></ul><ul><ul><li>Redundant controllers </li></ul></ul><ul><ul><li>Redundant cabeling to every node </li></ul></ul><ul><ul><li>Terminating on different HBAs </li></ul></ul><ul><ul><li>Storage based replication </li></ul></ul>
  14. 14. 3) Available solutions today <ul><li>Several implementations of Xen in the wild </li></ul><ul><ul><li>XenServer by Citrix </li></ul></ul><ul><ul><ul><li>Bundeled Hypervisor, out-of-the-box solution, including HA-mechanism </li></ul></ul></ul><ul><ul><ul><li>Version 5.5 released on June 16 </li></ul></ul></ul><ul><ul><ul><li>Free since Feburary 2009 </li></ul></ul></ul><ul><ul><li>Xen in Linux-Distributions </li></ul></ul><ul><ul><ul><li>different implementations of HA-mechanisms like Red Hat Cluster Suite </li></ul></ul></ul>
  15. 15. 4) Future prospect <ul><li>HA was only the first step </li></ul><ul><ul><li>Encapsulation is the key to more flexibility </li></ul></ul><ul><li>Fault-Tolerance is entering the market </li></ul><ul><ul><li>Xen Summit 2008: Demonstration of Kemari </li></ul></ul><ul><ul><li>Marathon everRun VM: Available </li></ul></ul><ul><ul><li>VMworld 2009 Europe: Announcement of VMware FT </li></ul></ul>
  16. 16. 4) Future prospect <ul><li>What is meant by FT? </li></ul><ul><ul><li>Virtual machine is running on two nodes simultaneously </li></ul></ul><ul><ul><li>Primary machine is visible to the outside world </li></ul></ul><ul><ul><li>Secondary machine is running as a shadow instance that can take over any time  no need for restarting the VM </li></ul></ul>Hardware Hardware Xen-Hypervisor Xen-Hypervisor VM 1 VM 2 VM 1 VM 2
  17. 17. 4) Future prospect <ul><li>FT-mechanisms are on the roadmap for Xen 4.0 </li></ul><ul><ul><li>Kemari (NTT, Japan): Unsupported version available </li></ul></ul><ul><ul><li>Remus (UBC, Canada): Initial port to Xen unstable (3.4.0) completed </li></ul></ul>
  18. 18. 5) Summary <ul><li>Virtualization brings more complexity to the infrastructure </li></ul><ul><ul><li>But: virtualization also brings more flexibility to the infrastructure </li></ul></ul><ul><ul><li>Obligatory redundant infrastructure for traditional cluster systems can be „recycled“ and used for multiple OS </li></ul></ul><ul><ul><li>High availability through encapsulation and hardware-independence of OS as a virtual machine </li></ul></ul><ul><ul><li>Fault-tolerance mechanisms are going to revolutionize availability of stand-alone-servers </li></ul></ul>
  19. 19. 6) Q&A <ul><li>- Any questions? </li></ul>
  20. 20. <ul><li>Thank you for your attention </li></ul><ul><li>For further questions feel free to contact me at </li></ul><ul><li>[email_address] </li></ul>

×