Standby Database Solutions for DB2 on i

Switched Disk

A switched-disk configuration uses disks that you can switch between different nodes in a high-availability cluster. You can use switched disk to create a simple and cost-effective high-availability solution for planned and some unplanned outages.

In a switched-disk environment, only one copy of the data exists, which means that performance is not impacted by data synchronization. In a switched-disk configuration the data is stored in an independent auxiliary storage pool (IASP). An auxiliary storage pool (ASP) is a collection of disks grouped together logically. To your system an ASP looks like a single unit of storage. An IASP is an ASP that can be located on internal or external storage, which you can bring online or take offline independent of system data or other ASPs. For more information about implementing SAP applications using an IASP, see SAP Note 568820 .

In a switched-disk environment, the distance between the production and backup system is limited by the physical length of the cables used to connect the systems. The maximum distance recommended is fifteen meters.

In addition to using IASPs, switched disk requires IBM i cluster technology. The cluster technology is shipped with operating system option 41 (HA Switchable Resources) .

You must configure an IASP as part of a high-availability cluster so the IASP is switchable.

Cross-site Mirroring (XSM)

Cross-site mirroring is a collective term for high-availability technologies that use different types of hardware replication. With hardware replication, a copy of the production data is mirrored to a backup system, so two or more copies of the data exist. The available variations of cross-site mirroring are geographic mirror, metro mirror, and global mirror.

Geographic Mirror

Geographic uses page-level mirroring performed at the operating system level. An exact copy of the production data is maintained on a backup system by utilizing synchronous mirroring. Synchronous mirroring means that the source system waits until acknowledgement from the target system that the data has been received. Synchronous mirroring keeps the data consistent and prevents data loss.

Geographic mirror lets you separate production and mirrored copies geographically. This offers protection in the event of a site-wide outage. However, due to synchronous communication, longer distances might impact performance and require more communication bandwidth. Both HA switchable resources (option 41 of the operating system) and IASPs are required for a geographic mirror solution. An external storage server is optional.

Metro Mirror

Metro mirror is similar to geographic mirror except that metro mirror uses mirroring at disk-sector level performed by an external storage server. In a metro-mirror environment the mirroring is synchronous, which prevents data loss between the production and backup system. Production data and backup data can be located on the same external storage server or on different external storage servers separated geographically. With separate storage servers the production and backup storage servers can be located up to 300 kilometers apart. However, due to synchronous communication, longer distances might impact performance and or require more communication bandwidth.

External storage servers, HA switchable resources (option 41 of the operating system), and IASPs are required for a metro-mirror solution.

Global Mirror

Global mirror uses disk I/O level mirroring between external storage servers. The mirroring is asynchronous, meaning that the source system does not wait for acknowledgement from the target system that data has been received. Depending on the distance between the storage servers, this means that the data on the backup system might lag the data on the production system by a few seconds. This allows the production and the backup server to be separated by an almost unlimited distance with no impact on application performance. External storage servers, HA switchable resources (option 41 of the operating system), and IASPs are required for a global mirror solution.

Logical Replication

Logical replication uses object-based replication performed at the middleware or application level. For journaled objects the data is replicated to a backup system by applying journal changes to the backup data. For non-journaled objects the objects are saved and then written to the backup system.

Since logical replication is performed at the object level, you can be more selective with data and object replication, which might reduce network traffic. The replication process can be synchronous to prevent data lag or asynchronous allowing almost unlimited distance between production and backup systems.

You can implement logical-replication environments using internal storage or external storage servers. Products that provide logical replication services are provided by IBM and independent software vendor (ISV) products. For more information about logical replication products, see the relevant website.