libvirt/docs/migration.html.in

633 lines
24 KiB
XML

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml">
<body>
<h1>Guest migration</h1>
<ul id="toc"></ul>
<p>
Migration of guests between hosts is a complicated problem with many possible
solutions, each with their own positive and negative points. For maximum
flexibility of both hypervisor integration, and administrator deployment,
libvirt implements several options for migration.
</p>
<h2><a id="transport">Network data transports</a></h2>
<p>
There are two options for the data transport used during migration, either
the hypervisor's own <strong>native</strong> transport, or <strong>tunnelled</strong>
over a libvirtd connection.
</p>
<h3><a id="transportnative">Hypervisor native transport</a></h3>
<p>
<em>Native</em> data transports may or may not support encryption, depending
on the hypervisor in question, but will typically have the lowest computational costs
by minimising the number of data copies involved. The native data transports will also
require extra hypervisor-specific network configuration steps by the administrator when
deploying a host. For some hypervisors, it might be necessary to open up a large range
of ports on the firewall to allow multiple concurrent migration operations.
</p>
<p>
<img class="diagram" src="migration-native.png" alt="Migration native path"/>
</p>
<h3><a id="transporttunnel">libvirt tunnelled transport</a></h3>
<p>
<em>Tunnelled</em> data transports will always be capable of strong encryption
since they are able to leverage the capabilities built in to the libvirt RPC protocol.
The downside of a tunnelled transport, however, is that there will be extra data copies
involved on both the source and destinations hosts as the data is moved between libvirtd
and the hypervisor. This is likely to be a more significant problem for guests with
very large RAM sizes, which dirty memory pages quickly. On the deployment side, tunnelled
transports do not require any extra network configuration over and above what's already
required for general libvirtd <a href="remote.html">remote access</a>, and there is only
need for a single port to be open on the firewall to support multiple concurrent
migration operations.
</p>
<p>
<img class="diagram" src="migration-tunnel.png" alt="Migration tunnel path"/>
</p>
<h2><a id="flow">Communication control paths/flows</a></h2>
<p>
Migration of virtual machines requires close co-ordination of the two
hosts involved, as well as the application invoking the migration,
which may be on the source, the destination, or a third host.
</p>
<h3><a id="flowmanageddirect">Managed direct migration</a></h3>
<p>
With <em>managed direct</em> migration, the libvirt client process
controls the various phases of migration. The client application must
be able to connect and authenticate with the libvirtd daemons on both
the source and destination hosts. There is no need for the two libvirtd
daemons to communicate with each other. If the client application
crashes, or otherwise loses its connection to libvirtd during the
migration process, an attempt will be made to abort the migration and
restart the guest CPUs on the source host. There may be scenarios
where this cannot be safely done, in which cases the guest will be
left paused on one or both of the hosts.
</p>
<p>
<img class="diagram" src="migration-managed-direct.png" alt="Migration direct, managed"/>
</p>
<h3><a id="flowpeer2peer">Managed peer to peer migration</a></h3>
<p>
With <em>peer to peer</em> migration, the libvirt client process only
talks to the libvirtd daemon on the source host. The source libvirtd
daemon controls the entire migration process itself, by directly
connecting the destination host libvirtd. If the client application crashes,
or otherwise loses its connection to libvirtd, the migration process
will continue uninterrupted until completion. Note that the
source libvirtd uses its own credentials (typically root) to
connect to the destination, rather than the credentials used
by the client to connect to the source; if these differ, it is
common to run into a situation where a client can connect to the
destination directly but the source cannot make the connection to
set up the peer-to-peer migration.
</p>
<p>
<img class="diagram" src="migration-managed-p2p.png" alt="Migration peer-to-peer"/>
</p>
<h3><a id="flowunmanageddirect">Unmanaged direct migration</a></h3>
<p>
With <em>unmanaged direct</em> migration, neither the libvirt client
or libvirtd daemon control the migration process. Control is instead
delegated to the hypervisor's over management services (if any). The
libvirt client merely initiates the migration via the hypervisor's
management layer. If the libvirt client or libvirtd crash, the
migration process will continue uninterrupted until completion.
</p>
<p>
<img class="diagram" src="migration-unmanaged-direct.png" alt="Migration direct, unmanaged"/>
</p>
<h2><a id="security">Data security</a></h2>
<p>
Since the migration data stream includes a complete copy of the guest
OS RAM, snooping of the migration data stream may allow compromise
of sensitive guest information. If the virtualization hosts have
multiple network interfaces, or if the network switches support
tagged VLANs, then it is very desirable to separate guest network
traffic from migration or management traffic.
</p>
<p>
In some scenarios, even a separate network for migration data may
not offer sufficient security. In this case it is possible to apply
encryption to the migration data stream. If the hypervisor does not
itself offer encryption, then the libvirt tunnelled migration
facility should be used.
</p>
<h2><a id="offline">Offline migration</a></h2>
<p>
Offline migration transfers inactive the definition of a domain
(which may or may not be active). After successful completion, the
domain remains in its current state on the source host and is defined
but inactive on the destination host. It's a bit more clever than
<code>virsh dumpxml</code> on source host followed by
<code>virsh define</code> on destination host, as offline migration
will run the pre-migration hook to update the domain XML on
destination host. Currently, copying non-shared storage or other file
based storages (e.g. UEFI variable storage) is not supported during
offline migration.
</p>
<h2><a id="uris">Migration URIs</a></h2>
<p>
Initiating a guest migration requires the client application to
specify up to three URIs, depending on the choice of control
flow and/or APIs used. The first URI is that of the libvirt
connection to the source host, where the virtual guest is
currently running. The second URI is that of the libvirt
connection to the destination host, where the virtual guest
will be moved to (and in peer-to-peer migrations, this is from
the perspective of the source, not the client). The third URI is
a hypervisor specific
URI used to control how the guest will be migrated. With
any managed migration flow, the first and second URIs are
compulsory, while the third URI is optional. With the
unmanaged direct migration mode, the first and third URIs are
compulsory and the second URI is not used.
</p>
<p>
Ordinarily management applications only need to care about the
first and second URIs, which are both in the normal libvirt
connection URI format. Libvirt will then automatically determine
the hypervisor specific URI, by looking up the target host's
configured hostname. There are a few scenarios where the management
application may wish to have direct control over the third URI.
</p>
<ol>
<li>The configured hostname is incorrect, or DNS is broken. If a
host has a hostname which will not resolve to match one of its
public IP addresses, then libvirt will generate an incorrect
URI. In this case the management application should specify the
hypervisor specific URI explicitly, using an IP address, or a
correct hostname.</li>
<li>The host has multiple network interfaces. If a host has multiple
network interfaces, it might be desirable for the migration data
stream to be sent over a specific interface for either security
or performance reasons. In this case the management application
should specify the hypervisor specific URI, using an IP address
associated with the network to be used.</li>
<li>The firewall restricts what ports are available. When libvirt
generates a migration URI it will pick a port number using hypervisor
specific rules. Some hypervisors only require a single port to be
open in the firewalls, while others require a whole range of port
numbers. In the latter case the management application may wish
to choose a specific port number outside the default range in order
to comply with local firewall policies.</li>
</ol>
<h2><a id="config">Configuration file handling</a></h2>
<p>
There are two types of virtual machines known to libvirt. A <em>transient</em>
guest only exists while it is running, and has no configuration file stored
on disk. A <em>persistent</em> guest maintains a configuration file on disk
even when it is not running.
</p>
<p>
By default, a migration operation will not attempt to modify any configuration
files that may be stored on either the source or destination host. It is the
administrator, or management application's, responsibility to manage distribution
of configuration files (if desired). It is important to note that the <code>/etc/libvirt</code>
directory <strong>MUST NEVER BE SHARED BETWEEN HOSTS</strong>. There are some
typical scenarios that might be applicable:
</p>
<ul>
<li>Centralized configuration files outside libvirt, in shared storage. A cluster
aware management application may maintain all the master guest configuration
files in a cluster filesystem. When attempting to start a guest, the config
will be read from the cluster FS and used to deploy a persistent guest.
For migration the configuration will need to be copied to the destination
host and removed on the original.
</li>
<li>Centralized configuration files outside libvirt, in a database. A data center
management application may not store configuration files at all. Instead it
may generate libvirt XML on the fly when a guest is booted. It will typically
use transient guests, and thus not have to consider configuration files during
migration.
</li>
<li>Distributed configuration inside libvirt. The configuration file for each
guest is copied to every host where the guest is able to run. Upon migration
the existing config merely needs to be updated with any changes.
</li>
<li>Ad-hoc configuration management inside libvirt. Each guest is tied to a
specific host and rarely migrated. When migration is required, the config
is moved from one host to the other.
</li>
</ul>
<p>
As mentioned above, libvirt will not modify configuration files during
migration by default. The <code>virsh</code> command has two flags to
influence this behaviour. The <code>--undefine-source</code> flag
will cause the configuration file to be removed on the source host
after a successful migration. The <code>--persist</code> flag will
cause a configuration file to be created on the destination host
after a successful migration. The following table summarizes the
configuration file handling in all possible state and flag
combinations.
</p>
<table>
<thead>
<tr class="head">
<th colspan="3">Before migration</th>
<th colspan="2">Flags</th>
<th colspan="3">After migration</th>
</tr>
<tr class="subhead">
<th>Source type</th>
<th>Source config</th>
<th>Dest config</th>
<th>--undefine-source</th>
<th>--persist</th>
<th>Dest type</th>
<th>Source config</th>
<th>Dest config</th>
</tr>
</thead>
<tbody>
<!-- src:N, dst:N -->
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="n">N</td>
<td>Transient</td>
<td class="n">N</td>
<td class="n">N</td>
</tr>
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="n">N</td>
<td>Transient</td>
<td class="n">N</td>
<td class="n">N</td>
</tr>
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y</td>
</tr>
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y</td>
</tr>
<!-- src:N, dst:Y -->
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="n">N</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y<br/>(unchanged dest config)</td>
</tr>
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="n">N</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y<br/>(unchanged dest config)</td>
</tr>
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y<br/>(replaced with source)</td>
</tr>
<tr>
<td>Transient</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y<br/>(replaced with source)</td>
</tr>
<!-- src:Y dst:N -->
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="n">N</td>
<td>Transient</td>
<td class="y">Y</td>
<td class="n">N</td>
</tr>
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="n">N</td>
<td>Transient</td>
<td class="n">N</td>
<td class="n">N</td>
</tr>
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="n">N</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="y">Y</td>
<td class="y">Y</td>
</tr>
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y</td>
</tr>
<!-- src:Y dst:Y -->
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="n">N</td>
<td>Persistent</td>
<td class="y">Y</td>
<td class="y">Y<br/>(unchanged dest config)</td>
</tr>
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="n">N</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y<br/>(unchanged dest config)</td>
</tr>
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="n">N</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="y">Y</td>
<td class="y">Y<br/>(replaced with source)</td>
</tr>
<tr>
<td>Persistent</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td class="y">Y</td>
<td>Persistent</td>
<td class="n">N</td>
<td class="y">Y<br/>(replaced with source)</td>
</tr>
</tbody>
</table>
<h2><a id="scenarios">Migration scenarios</a></h2>
<h3><a id="scenarionativedirect">Native migration, client to two libvirtd servers</a></h3>
<p>
At an API level this requires use of virDomainMigrate, without the
VIR_MIGRATE_PEER2PEER flag set. The destination libvirtd server
will automatically determine the native hypervisor URI for migration
based off the primary hostname. To force migration over an alternate
network interface the optional hypervisor specific URI must be provided
</p>
<pre>
syntax: virsh migrate GUESTNAME DEST-LIBVIRT-URI [HV-URI]
eg using default network interface
virsh migrate web1 qemu+ssh://desthost/system
virsh migrate web1 xen+tls://desthost/system
eg using secondary network interface
virsh migrate web1 qemu://desthost/system tcp://10.0.0.1/
virsh migrate web1 xen+tcp://desthost/system xenmigr:10.0.0.1/
</pre>
<p>
Supported by Xen, QEMU, VMware and VirtualBox drivers
</p>
<h3><a id="scenarionativepeer2peer">Native migration, client to and peer2peer between, two libvirtd servers</a></h3>
<p>
virDomainMigrate, with the VIR_MIGRATE_PEER2PEER flag set,
using the libvirt URI format for the 'uri' parameter. The
destination libvirtd server will automatically determine
the native hypervisor URI for migration, based off the
primary hostname. The optional uri parameter controls how
the source libvirtd connects to the destination libvirtd,
in case it is not accessible using the same address that
the client uses to connect to the destination, or a different
encryption/auth scheme is required. There is no
scope for forcing an alternative network interface for the
native migration data with this method.
</p>
<p>
This mode cannot be invoked from virsh
</p>
<p>
Supported by QEMU driver
</p>
<h3><a id="scenariotunnelpeer2peer1">Tunnelled migration, client and peer2peer between two libvirtd servers</a></h3>
<p>
virDomainMigrate, with the VIR_MIGRATE_PEER2PEER &amp; VIR_MIGRATE_TUNNELLED
flags set, using the libvirt URI format for the 'uri' parameter. The
destination libvirtd server will automatically determine
the native hypervisor URI for migration, based off the
primary hostname. The optional uri parameter controls how
the source libvirtd connects to the destination libvirtd,
in case it is not accessible using the same address that
the client uses to connect to the destination, or a different
encryption/auth scheme is required. The native hypervisor URI
format is not used at all.
</p>
<p>
This mode cannot be invoked from virsh
</p>
<p>
Supported by QEMU driver
</p>
<h3><a id="nativedirectunmanaged">Native migration, client to one libvirtd server</a></h3>
<p>
virDomainMigrateToURI, without the VIR_MIGRATE_PEER2PEER flag set,
using a hypervisor specific URI format for the 'uri' parameter.
There is no use or requirement for a destination libvirtd instance
at all. This is typically used when the hypervisor has its own
native management daemon available to handle incoming migration
attempts on the destination.
</p>
<pre>
syntax: virsh migrate GUESTNAME HV-URI
eg using same libvirt URI for all connections
virsh migrate --direct web1 xenmigr://desthost/
</pre>
<p>
Supported by Xen driver
</p>
<h3><a id="nativepeer2peer">Native migration, peer2peer between two libvirtd servers</a></h3>
<p>
virDomainMigrateToURI, with the VIR_MIGRATE_PEER2PEER flag set,
using the libvirt URI format for the 'uri' parameter. The
destination libvirtd server will automatically determine
the native hypervisor URI for migration, based off the
primary hostname. There is no scope for forcing an alternative
network interface for the native migration data with this
method. The destination URI must be reachable using the source
libvirtd credentials (which are not necessarily the same as the
credentials of the client in connecting to the source).
</p>
<pre>
syntax: virsh migrate GUESTNAME DEST-LIBVIRT-URI [ALT-DEST-LIBVIRT-URI]
eg using same libvirt URI for all connections
virsh migrate --p2p web1 qemu+ssh://desthost/system
eg using different libvirt URI auth scheme for peer2peer connections
virsh migrate --p2p web1 qemu+ssh://desthost/system qemu+tls:/desthost/system
eg using different libvirt URI hostname for peer2peer connections
virsh migrate --p2p web1 qemu+ssh://desthost/system qemu+ssh://10.0.0.1/system
</pre>
<p>
Supported by the QEMU driver
</p>
<h3><a id="scenariotunnelpeer2peer2">Tunnelled migration, peer2peer between two libvirtd servers</a></h3>
<p>
virDomainMigrateToURI, with the VIR_MIGRATE_PEER2PEER &amp; VIR_MIGRATE_TUNNELLED
flags set, using the libvirt URI format for the 'uri' parameter. The
destination libvirtd server will automatically determine
the native hypervisor URI for migration, based off the
primary hostname. The optional uri parameter controls how
the source libvirtd connects to the destination libvirtd,
in case it is not accessible using the same address that
the client uses to connect to the destination, or a different
encryption/auth scheme is required. The native hypervisor URI
format is not used at all. The destination URI must be
reachable using the source libvirtd credentials (which are not
necessarily the same as the credentials of the client in
connecting to the source).
</p>
<pre>
syntax: virsh migrate GUESTNAME DEST-LIBVIRT-URI [ALT-DEST-LIBVIRT-URI]
eg using same libvirt URI for all connections
virsh migrate --p2p --tunnelled web1 qemu+ssh://desthost/system
eg using different libvirt URI auth scheme for peer2peer connections
virsh migrate --p2p --tunnelled web1 qemu+ssh://desthost/system qemu+tls:/desthost/system
eg using different libvirt URI hostname for peer2peer connections
virsh migrate --p2p --tunnelled web1 qemu+ssh://desthost/system qemu+ssh://10.0.0.1/system
</pre>
<p>
Supported by QEMU driver
</p>
</body>
</html>