Roll Your Own Cloud: Enterprise Virtualization with KVM, DRBD, iSCSI and Pacemaker
|Day:||Wednesday 26 January 2011||Time:||13:30 - 14:15||Location:||L101|
|Project:||The Linux Cluster Stack: DRBD, Linux-HA/Heartbeat, Corosync, Pacemaker -- plus KVM, Libvirt, IET, and STGT|
A technical introduction to building a fully open source, hardware neutral, robust and efficient Enterprise Cloud. We present how to combine KVM, Libvirt, and the Pacemaker cluster manager to create feature-rich yet simple high availability for virtual machines. We then expand that concept to include shared-nothing and potentially split-site storage replication with DRBD. Finally, we demonstrate how to move to large enterprise configurations with large multiple-node clusters to provide highly scalable and flexible cloud computing capability.
The Linux Cluster stack, now in its third evolutionary iteration, makes an excellent basis for high availability on the Linux platform. The Pacemaker cluster manager is a highly flexible and feature-rich cluster resource management application. It currently interfaces with over 70 different server applications, Libvirt-based virtualization and iSCSI target services being just two of them. In this presentation, we will show how to build simple, yet fully redundant, replicated high availability cluster based on these components. We than gradually expand this concept to include scale-out to multiple cluster nodes, full live migration of virtual machines, and also centralized storage administration -- in other words, true enterprise-grade cloud infrastructure.
We focus exclusively on solutions that build on commercial, off-the shelf hardware and 100% open source technology.
Tim Serong, Melbourne-based Senior Clustering Engineer at Novell, will co-present.
Florian is a senior consultant, tech writer, blogger, and developer at
LINBIT. Working for a 100% Linux High Availability company, he blogs
about, and contributes to, Linux clustering development and storage
Tim is currently employed by Novell as Senior Clustering Engineer, working on the SUSE Linux Enterprise High Availability Extension. If HA clustering has taught him anything, it is that you can never ever have too many log files. Interests include motorcycling, beer, and making technology simpler and more reliable (although not necessarily in that order, and usually not all at the same time).