The Unified Archives feature is a tool for creating archives of physical or virtual systems to be used for system recovery and cloning operations. Virtual systems can be global zones, non-global zones, or kernel zones. Zones can be archived individually, bundled together, or selectively archived as a subset of the total zones. These multiple systems or instances are archived in a single unified format.
When you select zones to archive, by default all datasets associated with those zones are included. However, you can exclude any number of datasets from the archive.
A non-global zone archive contains the zone boot environment (BE) datasets and all delegated datasets assigned to it. A global zone archive, in addition, would also contain all datasets not associated with a zone. Entire top-down pools are also included.
Likewise, any fs resources in a zone are also excluded from the zone's archived data. However, an exemption exists in the case where the fs resource is defined as follows:
type=lofs special=path-to-a-ZFS-file-system
Because the special property refers to a directory for a ZFS file system, then an archive of the global zone would include the data in that directory.
With Unified Archives, you can perform the following:
Recover a system that needs to be replaced due to failure.
Duplicate or clone a system configuration that you want to install on multiple systems.
Migrate an existing system to new hardware or to a virtual machine.
An archive can either be a clone archive or a recovery archive.
A clone archive is the default archive type. It is based on the system's active BE when the archive is created. Therefore, only one BE can be archived at a time. You use a clone archive type for rapid creation and deployment of custom images.
If the clone archive contains both the global and non-global zones, the data for each archived system is stored independently and clone relationships are not preserved. Thus, each archived system is independently deployable. A cloned zone in a clone archive uses almost the same amount of space as its original zone.
A clone archive does not include any of the OS instance's system configuration information nor any sensitive data such as SSH keys or passwords. Any system deployed from a clone archive will have all of the file systems and applications as the cloned system, but not the same host name, for instance. System configuration information and sensitive data are preserved only in recovery archives.
You can create a dehydrated clone archive, where information from one or more publishers is removed, and then deploy it. See Example 5, Creating a Clone Archive without Content from Any Publishers and Example 6, Creating a Partially Dehydrated Clone Archive.
If the archive contains more than one deployable system, the publisher information is removed from all of the systems. If one of the systems does not include a named publisher, archive creation fails. You would need to create separate archives of those systems with their corresponding publishers.
You can rehydrate a dehydrated clone archive and then deploy it. Whether installing a dehydrated or rehydrated archive, you must use the same publisher with the same content version. Deploying a dehydrated clone archive takes longer due to the time to download information from the publishers.
A recovery archive is a full-system archive containing all BEs from all included instances. It consists of a single deployable system regardless of the combination of zones it contains. Unlike a clone archive, a recovery archive preserves zone clone relationships.
In a recovery archive, the active boot environment is the only fully prepared BE. Inactive BEs are archived mainly for data recovery regardless of their being bootable.
A recovery archive is intended for recovery operations, as in the case of hardware failure. Ideally, the archive should be deployed as part of a more comprehensive disaster recovery solution.
Unified Archives are compatible across hardware platform models of the same instruction set architecture (ISA). The archives are also portable to virtual machines on the same ISA. For example, an archive of an Oracle SPARC T5 system is deployable on the more modern Oracle SPARC M7 line of systems. In addition, such an archive is deployable on a logical domain or a Solaris kernel zone that runs on the same ISA.
Archive portability enables you to perform image transforms between physical and virtual machines as well as between global and non-global zones. Because these transforms are committed during deployment, the transform is determined by the deployed system type.
Provided that system ISAs are compatible, Unified Archives offers migration paths for legacy systems. For example, you can archive legacy hardware installations and deploy them as virtual machine instances on new hardware. Similarly, you can archive non-global Oracle Solaris zones and deploy them as Oracle Solaris kernel zones.
Unified Archives cannot be used in the following migration cases:
Oracle Solaris 10 branded zones cannot be transformed into any other system type. They may only be archived and deployed as an identical Oracle Solaris 10 branded zone instance.
Trusted Extensions or labeled zone archives can not be transformed. They can be deployed only to a global zone context, such as a Logical Domain, to Oracle Solaris kernel zones, or to bare metal machines.
Archives of labeled non-global zone archives are only deployable as 'labeled'-brand zones, running within the context of a Trusted Extensions global zone. See Planning Your Labeled Zones in Trusted Extensions in Trusted Extensions Configuration and Administration.