< Previous  | TOC | Index | Next  >

Creating a volume for use with snapshots and DRL

If the dirty region logging (DRL) or instant snapshot features are going to be used with an application volume, a version 20 data change object (DCO) and DCO volume must be associated with the volume.

If you want to create a volume of which you can take instant snapshots, the InstantSnapshottable capability or template must be specified (unless this capability is already implied by another capability or template), as shown in the following command:

# vxassist -g mydg -P mypool make mir3vol 1g \

  capability='DataMirroring(nmirs=3),InstantSnapshottable' \

  init=active

By default, a data change object (DCO) and DCO volume with a single plex are associated with a volume to enable the instant snapshot capability. To increase the redundancy of the DCO volume, specify the DCOLogMirroring capability as shown in the following command:

# vxassist -g mydg -P mypool make mir3vol 1g \

  capability='DataMirroring(nmirs=3),InstantSnapshottable,\

  DCOLogMirroring(nlogs=3)' init=active

In this example, the same degree of redundancy is created for the DCO volume as for the data volume.

The DCO and DCO volume that are added by the InstantSnapShottable capability also provide the dirty region logging (DRL) capability that speeds recovery of mirrored volumes after a system crash. To enable DRL, specify either the attribute drl=on for full DRL, or drl=sequential for sequential DRL (used with volumes containing database replay logs). By default, DRL is disabled (drl=off).

For example, the following command enables DRL for the volume, mir3vol, that is being created:

# vxassist -g mydg -P mypool make mir3vol 1g \

  capability='DataMirroring(nmirs=3),InstantSnapshottable' \

  drl=on init=active

Traditional DRL logs are implemented using a dedicated plex for each log copy, rather than using a DCO volume, and are supported for mirrored volume layouts only.

See Adding logs to a volume.