Theory of Operation & Deployment Choices

Theory of Operation & Deployment Choices

Root Disk Choices When Booting Nova Instances

There are several choices for how the root disk should be created which are presented to cloud users when booting Nova instances.

Boot from image

This option allows a user to specify an image from the Glance repository to copy into an ephemeral disk.

Boot from snapshot

This option allows a user to specify an instance snapshot to use as the root disk; the snapshot is copied into an ephemeral disk.

Boot from volume

This option allows a user to specify a Cinder volume (by name or UUID) that should be directly attached to the instance as the root disk; no copy is made into an ephemeral disk and any content stored in the volume is persistent.

Boot from image (create new volume)

This option allows a user to specify an image from the Glance repository to be copied into a persistent Cinder volume, which is subsequently attached as the root disk for the instance.

Boot from volume snapshot (create new volume)

This option allows a user to specify a Cinder volume snapshot (by name or UUID) that should be used as the root disk; the snapshot is copied into a new, persistent Cinder volume which is subsequently attached as the root disk for the instance.

Tip

Leveraging the “boot from volume”, “boot from image (creates a new volume)”, or “boot from volume snapshot (create new volume)” options in Nova normally results in volumes that are persistent beyond the life of a particular instance. However, you can select the “delete on terminate” option in combination with any of the aforementioned options to create an ephemeral volume while still leveraging the enhanced instance creation capabilities described in the section called “Theory of Operation: Enhanced Instance Creation”. This can provide a significantly faster provisioning and boot sequence than the normal way that ephemeral disks are provisioned, where a copy of the disk image is made from Glance to local storage on the hypervisor node where the instance resides.

Instance Snapshots vs. Cinder Snapshots

Instance snapshots allow you to take a point in time snapshot of the content of an instance’s disk. Instance snapshots can subsequently be used to create an image that can be stored in Glance which can be referenced upon subsequent boot requests.

While Cinder snapshots also allow you to take a point-in-time snapshot of the content of a disk, they are more flexible than instance snapshots. For example, you can use a Cinder snapshot as the content source for a new root disk for a new instance, or as a new auxiliary persistent volume that can be attached to an existing or new instance. For more information on Cinder snapshots, refer to the section called “Key Concepts”.

Instance Storage Options at the Hypervisor

The Nova configuration option instances_path specifies where instances are stored on the hypervisor’s disk. While this may normally point to locally attached storage (which could be desirable from a performance perspective), it prohibits the ability to support live migration of instances between hypervisors. By specifying a directory that is a mounted NFS export (from a NetApp FlexVol volume), it is possible to support live migration of instances because their root disks are on shared storage which can be accessed from multiple hypervisor nodes concurrently.

Note

Assuming shared storage (NFS) is used to store Nova instances, there are several other requirements that must be met in order to fully support live migration scenarios. More information can be found at http://docs.openstack.org/trunk/openstack-ops/content/compute_nodes.html

Warning

The NFS mount for the instances_path is not managed by Nova; therefore, you must use the standard Linux mechanisms (e.g. /etc/fstab or NFS automounter) to ensure that the NFS mount exists before Nova starts.