PowerHA Tools for IBM i - Full System FlashCopy

• On the production LPAR, pause database applications on a transaction boundary and flush pending writes to external disk

• Trigger the FlashCopy in the external storage unit

• Resume database activity on the production LPAR

• IPL the copy of the production LPAR and perform backups

• Optionally, transfer the BRMS data back to the source LPAR

• Ability to fully automate the power off or quiesce of the production system

• Creation of the FlashCopy ( warm flash/cold flash ) within an external storage unit

• Supports Storwize SVC (V3700, V5000, V7000, V9000) and DS8000 external storage

• Supports Storwize CSM

• Co-exist with HyperSwap and LPM

• Automatically power up the production system

• Power up the backup IBM i partition for save operations

• Provides exit point capability to submit backup jobs

• Provides integration with IBM’s BRMS

• Change host connections of the target partition

• Full system backup without ending user jobs

• Ability to have single command automation

• Integrates with SafeGuarded Copy to protect against ransomeware and similar

• Ease of Management utilities across multiple nodes

• Extensive logging

• Supported by phone (IBM Support 24x7) and email

The PowerHA Toolkit for IBM i version 4.6+ supports PowerVS API's. This allows it to operate in the IBM Cloud environment (it does not interface directly with PowerVC). While there are some key differences from the on-prem implementations of the toolkit the core functionality remains the same. The goal is to provide an automated backup process using Flash Copy to minimze outages to the production VM's.

A primary difference from the traditional (on-prem) toolkit implementations is the inability to directly access the storage and HMC. All of these interactions must be handled through the available PowerVS REST API's, which are layered on top of PowerVC. That removes some of the flexibility that we have on-prem and is reflected in the toolkit differences. Consider the following:

• The term 'volumes-clone' is used in the place of Flash Copy.
• The term VM or PVM is used in the place of LPAR.
• A VM must have at least one bootable volume attached.
• The volume marked as bootable has no relation to the IBM i load source.
• PowerVS does not expose VM adapters, SAN Zoning or tagged I/O to consumers.
• The maximum volumes attached to a VM is 127.
• PowerVS volumes-clone cannot re-use an existing target volume.

The toolkit takes all this into consideration and automates creating a usable copy of a production VM for backups. Compared to on-prem operations, the process of detaching, deleting, creating and attaching volumes takes additional time. Creating a snapshot (the basis of a volumes-clone) also takes longer than on-prem, so quiescing database operations might be an issue; rather than seconds, it can take minutes. *FRCWRT is always an option, and *IPL is not supported.

Similar to our on-prem model, a controlling VM with PowerHA Standard or Enterprise is required. The minimum requirements are the same as on-prem, with the exception that 0.25 CPU is the smallest increment avaialable.

The target (flashcopy) VMs must be provisioned and available; they are not deployed 'on the fly'. The toolkit does not automatically modify the resources (CPU, memory, etc) between flashcopies, but this can be done manually by customers. Nor does the toolkit support one target VM servicing multiple source VM volumes-clones. PTF SI80445 is required on all 7.4 IBM i VM's.

Toolkit licensing in PowerVS is slightly different than on-prem. It is licensed on a subscription model that includes entitlement to use the toolkit and access support. License keys are issued with expiration dates.

• Download the "What's new" attachment at the bottom of this page

• FlashSystem Policy-based volume group replication and flashcopy

• Uses REST API to communicate with FlashSystem instead of ssh

• Safeguarded Copy w/o CSM

• Ragged Commit Boundaries on FlashCopy

• Many improvements to PowerVS

• Many new useability features

• Production LPAR created and functional
  • Backup application installed and configured (including BRMS and iCC LPPs, which may require backup LPAR licensing)
  • Journaling is not required, but is strongly recommended
  • System value QALWUSRDMN must either be set to *ALL, or have QZRDHASM and QZRDMEDCLS libraries listed
    • In addition to the typical LPPs and PTFs required for the system to be usable, the following LPPs and PTFs should be installed:

      The following LPPs and PTFs are installed

      7.2	7.3	7.4	7.5
      5770WDS *BASE, 21	5770WDS *BASE, 21	5770WDS *BASE, 21	5770WDS *BASE, 21
      MF62565, MF64640, MF69797	MF62566, MF64641, SI77206, MF69809	MF69817, SI71911(SS1), SI72142(SS1), SI77407(SS1)	SI77521(SS1), TR1
      If using BRMS, the following PTF's are required (superseded is OK)
      SI71973(BR1), SI53860(SS1)	SI76737(BR1)	SI76738(BR1)
      If NOT using BRMS for backups, but 5770BR1 is installed, the following PTFs are required
      SI70366(BR1)	SI70367(BR1)	SI70368(BR1)

• Controlling LPAR created with the OS and PTFs installed
  • Can communicate (ping) the storage, HMC, and source LPAR
    • Controllers must be able to ping (iCMP) the HMC, Storage and production LPARs
    • Controllers use SSH (port 22) access the HMCs
    • Controllers use REST APIs and ssh (ports 22 and 7443) to the Flash System
    • Controllers use DSCLI (ports 1751 and 8542) to the DSs
    • Controllers must have public IP's to access PowerVS REST API endpoints
    • Controllers and production LPARs use port 55920 to communicate (this port is configurable)
    • PowerHA on the controllers must be able to communicate with each other on ports 5550 and 5551
  • Job CCSID must be 37
  • > 0.05 CPU, uncapped
  • 6 GB mainstore
  • > 400 GB DASD (enough for the OS, LPP's, PTF's etc)
  • In addition to the typical LPPs and PTFs required for the system to be usable, the following LPPs and PTFs should be installed:

    LPPs and PTF for the controlling LPAR

    7.2	7.3	7.4	7.5
    5733SC1 *Base, 1	5733SC1 *Base, 1	5733SC1 *Base, 1	5733SC1 *Base, 1
    5770SS1 30, 33, 41	5770SS1 30, 33, 41	5770SS1 30, 33, 41	5770SS1 30, 33, 41
    5770JV1, *Base, 16	5770JV1, *Base, 16	5770JV1, *Base, 16	5770JV1, *Base, 16
    5770HAS *Base and 1 or 2	5770HAS *Base and 1 or 2	5770HAS *Base and 1 or 2	5770HAS *Base and 1
    5770WDS *BASE, 21	5770WDS *BASE, 21	5770WDS *BASE, 21	5770WDS *BASE, 21
    Recent Group PTFs SF99776, SF99716 (level 27+)	Recent Group PTFs SF99876, SF99725	Recent Group PTFs SF99666, SF99665	Recent Group PTFs SF99676, SF99955
    5770HAS PTF SI57302, SI62180, SI65314
    5770999 PTF MF62565	5770999 PTF MF62566

• FlashCopy LPAR created, no OS is required (for compliance, all the cores must be licensed)
  • Recommend communications adapter is in a different slot (Cxx) than the adapter on the production LPAR
  • Available IP address
  • Sufficient CPU and memory to perform backups
  • Host connections to the FlashCopy LUNs
  • Necessary licenses for backup products (BRMS, iCC, etc)
• Backup Devices attached to all the LPARs (controllers, source and target)
  • Adapters and SAN Zoning
  • Libraries and volumes created
  • Enrolled in BRMS
• IP addresses, adminstrative user IDs and passwords for:
  • HMC's
  • LPAR's (including the FlashCopy target)
  • Storage devices (SVC / DS8K)
• The storage unit (DS8K/SVC) has:
  • Been installed and is running
  • FlashCopy licensed
  • Create LUNs for the target and RPF LPAR
  • Replication started (if using replication)
  • NOTE: DS firmware levels 7.8.30.470 or greater requires Java 8 (5770JV1 opt 16) on the controlling LPAR
  • NOTE: If using FS910 with GMCV and the change volumes are in a data reduction pool (DRP), the SVC must be at firmware level 8.2.1.1 or higher
• If using Recovery Point FlashCopy (RPF):
  • No target LPAR or IP addresses is required

• Read the joblog from the bottom up, and look for messages which indicate the problem. Many toolkit messages include recovery steps.

• Read the toolkit log (QZRDHASM/VIEWLOG) from the bottom up. Note that it may indicate information is provided in another log.

• Run the command QZRDHASM/DMPINF on the controller and provide the generated zip file to support.