IBM i Hosting i Environments

Positioning Information

PowerHA can be used in conjunction with an IBM i hosting i environment to provide disaster recovery for a production partition. This is not a standard PowerHA cluster/IASP environment, but rather a full system replication solution used by some customers as a DR solution. In this environment, the full production partition including the operating system is hosted in a network storage space within an IASP on the host partition. That IASP is then replicated using PowerHA to a backup host partition. In the event of a failure of the production partition, the target IASP and NWSD can be varied on, and the production partition can be IPLed on the second system.

In this environment, the production partition does not utilize an IASP, only the host partition has the IASP. However, there are several disadvantages of this technology compared to a traditional PowerHA environment where the production data only is within the IASP.

• There is no automatic failure detection of the production partition. A failure of the host partition would be detected, but no monitoring capability exists on the production partition.
• Since the entire production partition is being replicated, there is no ability to reduce OS upgrade time using this technology.
• For a planned switchover or a save from the target system, to be ensured a consistent copy, the production partition must be shut down before the target host and client can be varied on and IPLed.
• Before doing a system save from the target partition, the production system should again be shut down before detaching the IASP. A quiesce (going to restricted state or CHGASPACT) of the production system will help to reduce IPL time and also reduce the risk of damaged objects, but powering down the production system is the only way to guarantee that no partial objects are left in memory. Journaled objects should be recovered during the abnormal IPL of the target partition. Non-journaled objects may be marked damaged and would need to be recovered manually, just as in the case of a single system failure/recovery scenario.
• An unplanned failure of the production partition requires an abnormal IPL of the target, and may require some manual recovery of non-journaled objects.

As with all other solutions, journaling must be implemented so that the system can recover those journaled objects which may have been partially written to disk at the time of failure.

For planned operations (system save, switchover), in order to ensure a consistent copy on the target side (ie. no damage), the following steps must be taken.

To do a production system backup from the target side:

• Power down the production system
• Vary off the NWSD on the host system
• For a system save from the backup:
  • Vary off the IASP on the host system
  • Detach the IASP
  • Bring the production system back online by varying on the IASP and the NWSD on the host system and powering on the production system
  • Vary on the target IASP
• For a planned switchover:
  • Use the CHGCRGPRI or the PowerHA GUI to switch the IASP
• Vary on the NWSD on the target host system, and power on the target hosted system to do the system save

To do a planned switchover from the source production system to the target production system:

• Power down the production system
• Vary off the NWSD on the host system
• Issue CHGCRGPRI or use the PowerHA GUI to complete the switchover
• Vary on the NWSD on the new host system
• Power up the new production system

If the above steps are not followed, or in the case of an unplanned failure of the production system or of the host system, pages in memory may not have been written to disk on either the production or host system. When varying on the target IASP, non-journaled objects may be marked as damaged, and those objects would need to be recovered from a system save.

Setup Information

Here is a link to a document on how to set up an IBM i hosting IBM i environment. This document discusses many forms of virtualization available on IBM i. Chapter 2 is dedicated to IBM i hosting IBM i.

http://www-03.ibm.com/systems/resources/systems_i_Virtualization_Open_Storage.pdf

Here is a link explaining how to set up a PowerHA replication environment for a production server which is being hosted by IBM i. This document is an iSCSI solution guide, but the directions also pertain to an IBM i hosting IBM i environment. The relevant section of this guide for PowerHA implementations is section 7.4 - Implementing IBM i clustering for integrated servers.

systems_po_wer_ibmi_i_scsi_solut_ion_guide.pdf