A Practical Look at QEMU's Block Layer Primitives

  • Posted on: 29 September 2016
  • By: kashyapc
Track: 
Cloud Technologies
Day: 
Saturday
Author: 
Kashyap Chamarthy
Room: 
Track 3 (right)
Undefined
Paper: 

QEMU is an open source machine emulator and virtualizer (most commonly
used with Xen or KVM linux kernel module). In this talk, we'll focus on
QEMU's block subsystem, which forms the foundation for some of the
essential virtualization storage features -- live disk sychronization,
incremental backups, QCOW2 disk image chains, and point-in-time
snapshots to name a few. These features are driven by an underlying set
of block primitives, which are typically exposed to users via an
external virtualization API, such as libvirt.

We'll walk-through some of these primitives, discuss their invocation --
either directly using QEMU Machine Protocol (QMP) interface or the
libvirt APIs. We'll also understand how some of them could be combined
to perform specific useful operations, e.g. live disk backups; how live
storage migration is achieved via a combination of QEMU's built-in
Network Block Device server plus the live disk synchronization
mechanism. Higher-level cloud infrastructure such as OpenStack (its
Compute project) call into this QEMU block layer via libvirt APIs.

Audience
--------------

Audience would include Linux storage and Virtualization
(KVM/QEMU/Xen/libvirt) developers and admistrators, higher-level
infrastructure management software developers, or anyone interested
in the practical details of QEMU's block layer subsystem. Key take
away would be an understanding of some of the utilities QEMU's block
layer offers, and some practical details about how these can be used
to accomplish specific operations.

Time: 
14:00 - 15:00 hrs
field_vote: 
0
No votes yet