IBM i Functional Enhancements Details

November 2020 - IBM i 7.4 Technology Refresh 3 and IBM i 7.3 Technology Refresh 9 and IBM i 7.2

IPL Time Improvement - IBM i 7.4 and 7.3

November 2019 - IBM i 7.4 Technology Refresh 1 and IBM i 7.3 Technology Refresh 7

Enhanced Unmap Support for IBM Spectrum Virtualize

The SCSI unmap support in IBM i for storage based on Spectrum Virtualize is enhanced to allow the processing of the unmap commands to occur in the background. This can improve performance for systems that may have been driving a lot of SCSI unmap commands to the storage. A background task now frees the storage that has been deallocated (from an IBM i perspective) and allows for better management of data reduction pools.

Dynamic LUN expansion with the IBM DS8900
IBM i now supports the changing of the size of a LUN that comes from a DS8900 storage controller.  If the LUN type on a DS8900 is a variable LUN size (models 050 and 099), that LUN can increase in size.  The IBM i host might not fully recognize the size change immediately, but begins processing to prepare the internal structures for the size increase.  After the next IPL, the full capacity of the new LUN size is usable.  This can be used on all disks, including the load source.

May and June 2019 - IBM i 7.4

OS Limits increased for Max Disk Arms and LUN Size

Selected limits related to disk arms and LUN sizes have been increased in IBM i 7.4.  The new limits are shown in this table, along with the limit for IBM i 7.3 to use as a comparison:

Note 1 – Actual limit is one block short of the max shown in the table.

Note that for all 512/520 block disks/LUNs, the maximum size is still up to 2 TB.  This includes 5xx block SAS disks (internal or vSCSI attached), DS8000 LUNs, and IBM Storwize family LUNs.

For more information about these limits, and others, see IBM Documentation Miscellaneous Limits topic.

OS Limits Increased for Multi-core and multi-thread

The maximum numbers of cores varies by system model. Also, the operating system support of multi-core and multi-thread technology varies by operating system and release. The maximum number of processor cores and simultaneous multi-threading (SMT) levels supported by a single IBM i logical partition by processor mode are as follows:

• Supported maximums for IBM i 7.4 and IBM i 7.3:
  • POWER9®:  48 Cores / SMT8 / 384 Threads
  • POWER8®:  48 Cores / SMT8 / 384 Threads

IBM i 7.4 limits have been increased for the largest systems with POWER9 technology to enable up to 1536 threads to run, per partition, on up to 192 cores of SMT8.  Clients who require additional capacity beyond the IBM i supported maximum can contact IBM to request support for additional processor cores. IBM Systems Lab Services works with clients to assess the requirements and to provide IBM i support for the client workload in a partition with a larger number of cores.

September 2018 - IBM i 7.3 Technology Refresh 5 and IBM i 7.2 Technology Refresh 9

Unmap for IBM Spectrum Virtualize configurations

With IBM i 7.3 TR 5 and IBM i 7.2 TR 9, configurations attaching IBM Spectrum Virtualize is able to take advantage of the architected Unmap function. Both Native and VIOS NPIV configurations are supported, with or without a Fibre Channel switch.

Unmap can help manage disk utilization, especially for data reduction pools, but does require support on the external storage system also.  For the IBM i implementation, an Unmap is performed when an object is deleted, which informs the external storage system that the space is no longer needed by the Operating System.

There is no special action required to activate the IBM i support -- merely loading and applying the Technology Refresh PTF Group results in the function taking effect.

See more detailed information, including code level requirements for IBM Spectrum Virtualize, in IBM Support's topic SCSI Unmap Support in Spectrum Virtualize systems.

SAN Multipath Support for Tape Drives - Extended to additional tape drives

Back in November 2015, IBM i provided its initial support for SAN Multipath for tape drives with Automatic Fail-over.  (Refer to the November 2015 section below.) Multiple paths with automatic fail-over support reduce the chances of tape backups failing to complete within a backup window due to a SAN path failure.  The previous support included coverage for a large number of newer technology tape and tape library devices.  Now this support has been extended to also cover TS3500 (3584) with 3592-E05 and 3592-E06 drives.

Code requirements:

• IBM i 7.3 TR 5 (or PTF MF64645), or later
• Or IBM i 7.2 TR 9 (or PTF MF64644), or later

For more information on tape multipath support see: 

• IBM Documentation topic Tape Multipath
• developerWorks Wiki topic Using tape with multipath on IBM i 

March 2018 - IBM i 7.3 Technology Refresh 4 and IBM i 7.2 Technology Refresh 8

Unmap for IBM FlashSystems configurations

With IBM i 7.3 TR 4 and IBM i 7.2 TR 8, configurations natively attaching IBM FlashSystems 900 are able to take advantage of the Unmap function.  Unmap helps manage disk utilization and may help performance for both disk initialization and run-time.  There is no special action required to activate this support -- merely loading and applying the Technology Refresh PTF Group results in the function taking effect.

October 2017- IBM i 7.3 Technology Refresh 3 and IBM i 7.2 Technology Refresh 7

Increase in max LUNs per port for Fibre Channel configurations

Clients have been growing their configurations at a pace that has caused them to hit the limit on the number of LUNs they can create for each Fibre Channel port. The limit of 64 LUNs per port has now been increased to 127 for native-attached 16 Gb Fibre Channel adapters, and for Virtual Fibre Channel adapters (regardless of physical backing adapter).

IBM i system back-up saves LAN console configuration as part of the LIC data

IBM i 7.3 TR3 and IBM i 7.2 TR7 include recently released PTFs that eliminate steps when installing IBM i onto a new load source from a SAVSYS. The LAN console configuration is now automatically saved to a new LIC LID during a SAVSYS operation. This enables the LAN console server to start during the D-mode IPL and be ready for the IBM i Access Client Solutions (ACS) connection. There is no longer a need to pull together the LAN console parameters and configure it separately if the same LAN console is being used for the new load source installation.

No user actions are required to take advantage of this new support. The system automatically attempts to use the previous configuration during a D-mode IPL that uses a backup created from a partition that had IBM i 7.3 TR3, or later, or IBM i 7.2 TR7, or later, code level.

Automatic sorting of volumes in optical container media
The volumes on an optical container media used for installation are now automatically sorted, so there is no longer a need to create a VOLUME_LIST file or to run VFYIMGCLG to first get the volumes sorted in proper order for installation.

The operating system also verifies that there is only one bootable media in the catalog; any subsequent bootable volumes are dropped from the loaded list and a LIC log is generated.

Maximum disk and load source size increase to 4 TB for 4160-byte sectors

The maximum size for an individual disk, and for the load source, in an IBM i partition has increased from 2 TB to 4 TB for 4160-byte sector disk devices.  4160-byte sectors are used by the newer direct attach disk storage devices, both natively and in some VIOS configurations.  Note that load sources with 4096-, 520-, and 512-byte sectors remain capped at 2 TB.

No maximum for USB flash drive capacity
USB flash drives with a capacity greater than 32 TB are now allowed for an IBM i partition.

Improved debug for storage IOAs
Changes have been made in the storage I/O infrastructure to improve debug capabilities for storage IOA failures.

Improved performance for some IBM DS8000 storage configurations
Some configurations that include IBM DS8000 storage systems may get a performance benefit due to changes in the implementation of low-level I/O architecture.

March 2017- IBM i 7.3 Technology Refresh 2 and IBM i 7.2 Technology Refresh 6

Serviceability improvements for severe errors

Focus has been placed on the disruption to production that occurs when a severe error situation requires a main storage dump analysis by IBM. Serviceability improvements with IBM i 7.3 TR 2 and IBM i 7.2 TR 6 preclude the need to take a main storage dump for some error scenarios, allow faster copying of main storage dumps to tape and to Integrated File System (IFS), and significantly reduce the time to load a main storage dump for analysis.  

More specifically, there are some error situations where a logical partition (LPAR) is still running, but in a degraded mode.  If a console display or workstation display can be reached, then a standard collection of service information can be sent to a series of LIC logs for later analysis by IBM.  With IBM i 7.3 TR 2 or IBM i 7.2 TR 6, if the console display or workstation display cannot be started, then a new operator panel sequence (Function 21 followed by another Function 21) can be used instead.  This sequence captures that same standard collection of service information and place it into a series of LIC logs.  The partition might still need to be restarted to correct the error situation, but the time for creating and saving a main storage dump is avoided.  The LIC logs can be sent to IBM for analysis at the earliest convenient time.

Note that with this enhancement, when Function 21 is started on the op panel,  SRC A60050A0 0D200060 00000000 is displayed.  When a second Function 21 is started, and the display fails to activate, and it is within 5 minutes of the first Function 21, then SRC A60050A0  0D200060 11110000 is displayed. This means the "log servicedoc" function is being called.  There are multiple SRCs displayed as the log function progresses.  Those steps are denoted with SRCs in the format 0D200060 mmmmnnnn, where mmmm indicates which step is running, and nnnn indicates the total number of steps.  When the log function completes, then SRC A60050A0 0D200060 CCCC0000  is displayed.

For error situations that still require a main storage dump, the copy times for copying that dump to IFS and to tape have been significantly reduced.  Whether being viewed on a customer's system or on a system owned by IBM Service, the time required to format a main storage dump for analysis has also been greatly reduced.  Together these improvements result in faster analysis and action to resolve serious problems.

Flash drive no longer requires a physical pull or push when exported

An exported USB flash drive can now be re-imported via the use of a command sequence, without needing a physical pull or push of the flash drive.

Default setting for 4096 sector disk has been changed
The default setting for the cleaning of "dirty free space" for 4096 sector disk has been changed.
DS8000 HyperSwap disks no longer require resume of normal replication before IPLing
Systems with DS8000 HyperSwap disk units configured can be successfully IPLed following an unplanned fail-over, even if normal HyperSwap replication was not resumed before the IPL.

November 2016 - IBM i 7.3 Technology Refresh 1 and IBM i 7.2 Technology Refresh 5

Containers make USB Flash and RDX look like a stack of DVDs

Clients have been looking for more operations that can be done with USB flash drives and to make better use of the large-capacity RDX media. Today, clients can create optical image catalogs only on disk (IFS). With IBM i 7.3 TR1 and IBM i 7.2 TR5, Container support has been added so that clients can now store an image catalog into the QOPT file system on a USB removable flash drive or RDX media, making the format of the data on the flash drive or RDX media look like a stack of DVDs. This new Container support can ease operational procedures in many ways:

• A system administrator can download an ISO image that contains PTFs onto a PC at the office, move that file to a USB removable flash drive, and apply the PTFs directly from the flash drive. Prior to this Technology Refresh, you simply could not use a flash drive to load PTFs. It was necessary to store the image on disk within an IBM i partition and then either add it to an Image Catalog or burn it to DVD.
• A system administrator can now store an entire Image Catalog on a USB removable flash drive or RDX media. Previously, Image Catalogs were limited to configurations with DASD and virtual media, but this is more expensive than USB flash drives and RDX media. Putting images on a USB flash drive also makes it easy to move to another system, which makes it easy for a business partner or ISV to take with them to a customer site.
• With up to 2 TB of capacity, a system administrator could now consider partitioning an RDX media in a way that enables multiple incremental saves, such as an incremental save for each day of the week for nightly backups, or each week of the month for weekly backups, and so on. Previously, if incremental saves were done to separate directories on an RDX media, and the contents of the directories were deleted to enable reuse, file fragmentation on the media could cause noticeable performance reductions. Now the media can be subdivided into multiple virtual disks to be used for saves. Then anytime one of those virtual disks is reinitialized, there is no fragmentation of the data. If the save operation creates a bootable disk image, it does not mean that this media can be used for a D-mode IPL, because the device is a virtual disk, and virtual disks in general cannot be used for a D-mode IPL. This image is intended to be used for restore operations after the partition is IPLed.

For more details on usage and restrictions, see Removable Media on IBM i.

New format for tape volume statistics

A new format for tape volume statistics has been introduced in IBM i 7.3 TR1 to store larger read/write counts. This cuts down on the number of Product Activity Log entries for statistics that are being generated, thereby reducing the frequency of wrapping the log and allowing applications, such as BRMS, to properly track cartridge usage.

Identifying a partition with a UUID

Power Systems Firmware V8.2 and later generates a 16-byte Universally Unique Identifier (UUID) as an attribute for each logical partition (LPAR). With IBM i 7.3 TR1 and IBM i 7.2 TR5, this UUID can then be materialized using the Materialize Machine Attributes (MATMATR) instruction. The UUID does not change when the LPAR moves to a new serial number; for example, during a Live Partition Mobility (LPM) operation. While the IBM i operating system, the Licensed Program Products for IBM i, and the IBM i license management function do not utilize the UUID, application providers may choose to utilize the UUID in order to facilitate movement and management of LPARs between servers. For information regarding using the machine instructions related to UUID, refer to Machine Interface instruction MATMATR in IBM Documentation, and search for "UUID."

Here is a simple coding example, which prints the LPAR UUID in a hex digit format (e.g., 01234567 89ABCDEF 01234567 89ABCDEF):

   _MMTR_Template_T matmatr0204;
   memset(&matmatr0204, 0, sizeof(matmatr0204));
   matmatr0204.Options.Template_Size = sizeof(matmatr0204);
   short mmtrAttribute = _MMTR_UUID;
   _MATMATR1(&matmatr0204, &mmtrAttribute);

   printf("%08X %08X %08X %08X\n",
       * (unsigned long *) &matmatr0204.Options.Data.PartitionUUID[0],
       * (unsigned long *) &matmatr0204.Options.Data.PartitionUUID[4],
       * (unsigned long *) &matmatr0204.Options.Data.PartitionUUID[8],
       * (unsigned long *) &matmatr0204.Options.Data.PartitionUUID[12]);

April and May 2016 - IBM i 7.3 and IBM i 7.2 Technology Refresh 4

SSD Fuel Gauge

IBM i has added support for Read Intensive Solid State Drives (SSDs).  These drives are intended for workloads that are not heave on writes, since the drives have limited write capability.  Along with that support, there is a new function for monitoring the number of writes to these Read Intensive SSDs.  See IBM i SSD Fuel Gauge for more information.

November 2015 - IBM i 7.2 Technology Refresh 3
DUPOPT of IPL capable media enhancement
Support has been added to IBM i 7.2 TR3 to duplicate IPL-capable optical media to optical media with different physical characteristics. Specifying a special value of *BOOT on the DUPOPT "To volume identifier" (TOVOL) parameter indicates that the boot area and all files from the source media are to be copied to the mounted target media, even if the physical characteristics of the target optical volume do not match the physical characteristics of the source optical volume. This process enables duplicating an existing IPL-capable DVD or distributed media image to RDX or flash media. System administrators find this useful when transitioning to the use of newer media types. ISVs find this especially convenient for condensing a bootable install image onto a flash drive that can be carried in one's pocket.

See IBM Documentation topic Duplicating optical volumes for more information.

SAN Multipath Support for Tape Drives - Automatic Fail-over

Back in May 2015, IBM i provided its initial support for SAN Multipath for tape drives.  (Refer to the May 2015 section.)  That initial support required manual Vary Off and Vary On when failures occurred.  Now with IBM i 7.2 TR 3, the SAN Multipath configurations can be automatically recovered for most types of errors; the operating system just switches the I/O to an alternate path.  In any cases where automatic recovery is not possible, the manual fail-over support added with IBM i 7.2 TR2 can still be used.  In either case, the new path remains the active path until a failure occurs on that path, or until the configuration is varied off and back on.  Automatic fail-overs reduce the chances of tape backups failing to complete within the backup window due to a SAN path failure.

HDD RAID array start time reduced
Those who set up new IBM i partitions with local SAS Hard Disk Drives (HDDs) benefit from support that has been added to the newer SAS Input Output Adapters (IOAs) and IBM i.  Especially as disk capacities have increased, the time it takes to start a RAID array for HDDs has also been increasing.  Support has been added so that for SAS HDD disk configurations, the time it takes to start a RAID array is more proportional to the amount of data on the drives.  This can significantly reduce the time to start the array for some configurations.

IBM i supports SVC/Storwize HyperSwap Volumes
IBM i 7.2 TR3 now supports the HyperSwap function of the IBM SAN Volume Controller (SVC) and IBM Storwize line of products. The HyperSwap function enables highly available disk volumes accessible through two storage controllers.  These controllers can be in the same data center, or located kilometers or miles apart.   At this point in time, IBM  i PowerHA SystemMirror does not support HyperSwap on SVC or Storwize. For more information, see IBM Spectrum Virtualize Software V7.5 announcement letter from May 2015.

May 2015 - IBM i 7.2 Technology Refresh 2, IBM i 7.1 Technology Refresh 10

SAN Multipath Support for Tape Drives (IBM i 7.2 TR 2 only)

With IBM® i 7.2 TR 2 and later code levels, support is added to allow most Fibre Channel tape and tape library devices to be attached multiple times to the same IBM i partition.

• Up to eight paths are supported for each device.
• The multipath functions available depend upon the capabilities of the attached tape and tape library devices.
• For maximum redundancy, you can use both ports of the tape devices to attach to the partition.
• Because tape devices are sequential access devices, only one path to the device can be active at a time.
• Attachment may be native to the IBM i partition, or via a VIOS NPIV configuration, or a mix.

For the newer technology tape and tape library devices, attaching the same device multiple times to different adapter or adapter ports on the same the IBM i partition provides a configuration where I/O path failures can be recovered by vary off and vary on of the affected device descriptions, or deallocate and allocate of a failing resource within a tape library device, to switch to an alternative path.

Multipath support with manual fail-over is provided for the following tape and tape library devices:

• Fibre Channel LTO5 and newer drives in the 7226 enclosure
• TS3100/TS3200 (3573) with LTO5 and newer Fibre Channel drives
• TS3310 (3576) with LTO5 and newer Fibre Channel drives
• TS3500/TS4500 with LTO5 and newer, and 3592-E07 and newer Fibre Channel drives
• ProtecTIER® virtual tape library, code level 3.3.5.1 or newer

Note: The device driver uses persistent reservation for drives that are supported with multipath.  If a device is attached that emulates a supported IBM tape drive, but does not support persistent reservation as expected, the results may be unpredictable.

With this new function, for most of the older technology tape and tape library devices (not mentioned in the above list of supported drives), attaching the same device multiple times to a different adapter or adapter ports on the same IBM i partition provides a usable configuration that behaves the same as if there was only one path to the device for most operations, but does not support recovery from path failures.

Multipath tape attachment is not supported at all for the following devices.

• SCSI attached tape devices.
• 3494 tape library.

See the IBM Documentation topic for Tape Multipath for additional information.

SAS Adapter Availability Improvement

Innovative improvements have been made in IBM i 7.2 TR 2 and IBM i 7.1 TR 10 to drastically reduce the recovery time for some recovery and maintenance scenarios for SAS dual adapter configurations. The breadth of impact for some of the scenarios has been greatly reduced also, to not affect the operation of active paths.  These improvements apply to the following adapters and controllers:

• #EJ0L - PCIe3 12 GB Cache RAID SAS Adapter
• #EJ0N, #EJ0P - PCIe3 integrated SAS controller in Power S814 and S824 servers
• #EJ0J, #EJ0M - PCIe3 RAID SAS Adapter with Zero Write Cache

Examples of scenarios that benefit from this enhancement include:

• Customers with SAS dual adapter configurations that use system level mirroring and encounter some types of adapter and disk failures avoid the temporary performance degradation and/or a possible temporary hang condition that might have lasted many minutes before this improvement.

• Customers with RAID-protected dual adapter configurations who encounter similar temporary performance degradations from some types of adapter errors or when performing adapter concurrent maintenance operations.

• An SSR replaces a drive as a result of a predictive error for a mirrored disk drive.  The SSR replaces the drive using disk concurrent maintenance and can pull the drive without affecting the mirroring performance for an extensive amount of time.

Previously these scenarios resulted in a reset and reload of both adapters in a dual adapter configuration.  With this improvement, only the adapter that requires a recovery action is reset and reloaded.  Meanwhile the other adapter automatically takes over the active paths that were assigned to the resetting adapter.  After the reset and reload of the one adapter, the active paths that were previously assigned to it are moved back to its control.

November 2014 - IBM i 7.2 Technology Refresh 1

Decreased Minimum Load Source size
Back in April 2014, IBM announced that the new minimum load source size for IBM i 7.2 partitions would be 70 GB of usable space.  Now in October 2014 IBM is announcing that although native SAS and SCSI configurations still require a minimum of 70 GB usable space for a load source, virtual and external storage LUN configurations can use a load source that is a minimum of 35 GB usable space.  Note that PTFs are required on an IBM i 6.1 or 7.1 partition before upgrading to IBM i 7.2. 

For details about calculating "usable space" and about the PTFs required, see the developerWorks topic IBM i and Related Software.

May 2014 - IBM i 7.2

IBM i support for "WAN over LAN" configurations

IBM i 7.2 includes new support for "WAN over LAN" configurations to expand advanced virtualization capabilities:

• Saves PCI slots for applications requiring a modem, thus reducing configuration cost
• Provides WAN connectivity for servers that do not have available PCIe slots
• Allows IBM i client partitions with virtual I/O to use FAX and other WAN applications
• Allows the same Ethernet adapter that provides TCP/IP connectivity for an IBM i partition to provide WAN connectivity

IBM i applications that have traditionally run on PCIe WAN adapters now run seamlessly over an Ethernet connection to an Ethernet device server without application changes. For an existing IBM i configuration definition, the only change is that the Async or PPP line description must be modified to specify that it is for an Ethernet device server, and specify the Ethernet device server IP address and port. IBM i clients requiring a large number of serial ports or running on systems without PCIe adapter slots (such as PureSystems™ or BladeCenter® devices) now have the option to run IBM i Communication Applications on their systems.

IBM i on Power Systems has been providing WAN communications via system adapters. The #5289 - 2 Port Async EIA-232 PCIe / #5290 - PCIe LP 2-Port Async EIA-232 Adapter provides an RS-232 interface for two external devices, such as modems. The #2893 - PCIe 2-Line WAN w/Modem adapter / #2894 - PCIe 2-Line WAN w/Modem CIM adapter provides an integrated modem in addition to a single RS-232 interface. Now IBM i also supports Ethernet device servers, which provide multiple RS-232 serial ports. Ethernet device servers, which are also known as Ethernet serial servers or Ethernet terminal servers, are external non-IBM equipment that attaches to an Ethernet and contains one or more serial ports to which external modems can be attached. An Ethernet device server is accessed over the network using a standard TCP/IP sockets connection. Thus, IBM i now has true virtual serial ports for applications, such as IBM Facsimile Support for i, 5798-FAX.

IBM i requires an Ethernet device server to support the Telnet RFC 2217 COM Port Control Extension for transport of serial data and RS-232 control signals. Ethernet device servers with RFC 2217 support are available from a variety of vendors, in models containing anywhere from 1 to 48 serial ports. IBM i 7.2 has been tested and found to work with the Digi PortServer TS 4 (firmware version 82000747_V2 01/29/2013) and Perle IOLAN SDS4 (firmware version 4.5) Ethernet device servers. Facsimile Support for IBM i has been tested with the MT9234ZBA modem attached to these Ethernet device servers.

IBM i support is limited to IBM's implementation of Telnet RFC 2217 COM Port Control Extension and IBM software and hardware. The non-IBM vendor provides the support for its implementation of RFC 2217, its TCP/IP network connection, and any downstream modems or fax devices. In case a problem does occur at the client site, like other non-IBM products attached through IBM LAN adapters and standard protocols, IBM participates with clients and other vendors in problem determination under standard IBM warranty and service agreements. If IBM determines the problem is associated with the IBM adapter or with the IBM protocol implementation, normal problem resolution procedures follow.

See additional details at these links:  

• WAN over LAN Connection Scenario
• CRTLINASC
• CHGLINASC
• CRTLINPPP
• CHGLINPPP
• IBM Facsimile Support for IBM i 
• Modem Multitech MT9234ZBA
• Ethernet Device Servers from Digi
• Ethernet Device Servers from Perle

Increase in maximum file size for IFS QOPT files

Prior to IBM i 7.2, the maximum file size supported by the /QOPT file system was about 4 GB, which worked well for accommodating the size of DVD media. With 7.2 the maximum size is increased to almost 10 TB. This better accommodates the size of newer media types, such as the recently introduced RDX media that are over 1 TB in size. With this larger file size, it is possible for languages and utilities to access large files and perform operations, such as copying IBM i distribution images to RDX media to move them to another partition or system. For more specifics visit File size for IFS QOPT files in IBM Documentation.

Improved status display for aggregated Ethernet ports

With IBM i 7.2 Ethernet line descriptions now display more detailed status information for aggregated ports, making it easier to diagnose and correct any issues with performance or functionality.  See IBM Documentation topic Ethernet Link aggregation for more information about this technology.

Usability enhancements for vary on of tape and tape library devices

With IBM i 7.2 Vary On processing is improved for tape and tape library devices.  It is no longer necessary to set the device description "Online at IPL" parameter to *NO and use a startup program to vary them on after the IPL. The device description "Online at IPL" parameter can now be set to *YES to allow the tape and tape library devices to be varied on and available after a system IPL.

The amount of time to wait for resources to be allocated during vary on processing for tape libraries has been changed to increase as the number of resources in the tape library is increased. This should reduce the number of "The request did not complete in the time allotted" failures when varying on or changing the resource allocations for tape libraries.

The amount of time to wait for a vary on with reset to complete has also been increased. This should reduce the number of vary on with reset failures that need to be recovered by another vary on attempt without the reset.

November 2013 - IBM i 7.1 Technology Refresh 7

LACP for Aggregate Ethernet Line Descriptions

IBM i 7.1 Technology Refresh 7 has added Link Aggregation Control Protocol (LACP) for Aggregate Ethernet Line Descriptions to allow for easier configuration.  Specifically, it allows simpler switch configuration and detects configuration problems.

The Create Line Desc (Ethernet) (CRTLINETH) and Change Line Desc (Ethernet) (CHGLINETH) commands are used to manage Ethernet line descriptions, including aggregate line descriptions (indicated by Resource name (RSRCNAME) *AGG).  For an aggregate line description, the Aggregate policy (AGGPCY) has two elements: "Standard", which controls negotiation with the link partner, usually a switch, and "Policy type", which controls which Ethernet port is used to send each outgoing frame.

Prior to IBM i 7.1 Technology Refresh 7, only static aggregation was supported, so the main documentation says that the standard must be set to *ETHCHL.  With Technology Refresh 7, IBM i adds support for the Link Aggregation Control Protocol (LACP) via the *LNKAGG option.  For environments that support it, LACP offers two key benefits: detection and tolerance of incorrect configurations (like invalid cabling or link partner settings) and easier switch configuration.

Choosing *ETHCHL as the Standard uses static aggregation, which performs no negotiation with the link partner.  The link partner must also be configured for static aggregation.  The Ethernet ports in the Aggregated Resource List (AGGRSCL) and the set of ports configured at the link partner should correspond exactly.  If they do not, then some Ethernet packets may not get to the correct destination.  Note that there are several opportunities for error: having ports in AGGRSCL that are connected to a different link partner or to the wrong ports in the correct link partner, or having ports selected in the link partner's aggregate that are connected to ports not in AGGRSCL.

Setting the Standard as *LNKAGG instead forces use of the Link Aggregation Control Protocol (LACP) as described in the IEEE 802.3ad standard.  This negotiation detects the identity of the link partner for each Ethernet port in the Aggregated Resource List (AGGRSCL).  This requires that the link partner enable LACP with common identification information on all ports that are connected to ports in AGGRSCL.  In order for a port to aggregate and be used for Ethernet traffic, its partner port must respond to the LACP negotiation, and the response must match the identifying information to all other aggregated ports.  The aggregation panel in Display Line Description (DSPLIND) shows the aggregation status of each port.  For any port that has not joined the aggregate, the status should help determine why.

June 2013 - IBM i 7.1 Technology Refresh 6 and IBM i 6.1

Variable LUN Sizes for DS8870 Storage Systems

IBM DS8870 now supports variable LUN sizes for IBM i with its new R7.1 release, adding flexibility in sizes and potentially increasing utilization of the storage system.  Previously the DS8000 storage systems have been restricted to concrete sizes of LUNs, such as 17.5 GB, 35.1 GB, etc.  With this June code update, one can create variable LUN sizes for IBM i partitions, in increments ranging from 1 GB up to 2000 GB.  The sizes must be in full GB increments.  Support is provided with IBM i 7.1 TR 6 PTF Group or IBM i 6.1 Resave RS-611-K, or later.  See the IBM System Storage DS8870 RFA for more information.

Usage note:  Dynamic resizing of existing LUNs is not included in this new support -- the wider range of sizes is available only at creation time.

February 2013 - IBM i 7.1 Technology Refresh 6 and IBM i 6.1

Ethernet Link Aggregation Enhancements

IBM i support for Ethernet Link Aggregation was first introduced with IBM i 7.1 TR 3; see that announce information in the October 2011 section. That function is being enhanced for IBM i 7.1 TR 6. As before, Ethernet ports that successfully established their link when the line description was varied on, but then lost and re-established their link, rejoin the aggregate. However, there is now more automatic recovery from hardware failures for individual Ethernet resources in an aggregate:

Ethernet ports that fail to establish link when the line description is varied on now periodically attempt to rejoin the aggregate, as do ports that encounter errors other than lost link.

Also, a message with ID CPPEA69 is displayed whenever an Ethernet resource joins or leaves an aggregate. A user can monitor for that message to allow for more decisions and automation. The message indicates which resource was affected, whether it joined or left the aggregate, and how many resources are currently aggregated.

Performance Improvement to LIC PTF Apply

Apply time has been improved for both PERM and TEMP apply LIC PTFs with IBM i 7.1 TR 6 and IBM i 6.1 on some configurations. This particular benefit is likely to be noticed mainly on smaller partitions and systems because they are more likely to be constrained in memory and processing power, though individual results vary. Partitions and systems that are not constrained in memory or processing power are not likely to see any noticeable benefit.

This enhancement is part of the IBM i 7.1 TR 6 PTF Group. If you have a configuration that you believe would see some potential performance benefit, and you already have IBM i 7.1 TR 5 PTF Group on your partition, you would see that benefit during the apply of the IBM i 7.1 TR 6 PTF Group. If you do not already have IBM i 7.1 TR 5 PTF Group on your system, you would experience the benefit with subsequent PTF applies.

This enhancement is also being included in the HIPER PTF Groups and PTF CUM packages for IBM i 7.1 and IBM i 6.1. You can check for these PTFs on your system to see if you already have them from one of these sources:

• IBM i 7.1 PTF MF56423
• IBM i 6.1 PTF MF45484

October 2012 - IBM i 7.1 Technology Refresh 5 and IBM i 6.1

Enhanced Administrative SMT controls for POWER7

The POWER7 processor supports up to 4-way Simultaneous Multi-Threading (SMT) with Intelligent Threads, whereby the hardware and operating system cooperate to maximize workload throughput regardless of the number of hardware threads required by the workload. By default, IBM i enables SMT4 for POWER7 when processor multitasking (system value QPRCMLTTSK) is system controlled or explicitly enabled. When SMT4 is enabled, the operating system uses up to four hardware threads per processor as demanded by the workload. When the workload demands more than four hardware threads per processor, it is queued and subject to operating system scheduling policy. In most cases, the default SMT control values provide the best balance between single thread performance and maximum system throughput.

The enhanced SMT controls in IBM i 7.1 enable the system administrator to manually limit the number of hardware threads per processor used by the operating system when processor multi-tasking is enabled:

• The Change processor multitasking (QWCCHGPR) API can be used to set the maximum number of secondary hardware threads per processor on systems that support simultaneous multithreading. For POWER7, '0' denotes no maximum, '1' denotes SMT2, and all other values denote SMT4.
• The Retrieve Processor Multitasking Information (QWCRRTVPR) API can be used to retrieve the maximum number of secondary hardware threads per processor on systems that support simultaneous multithreading.

A change to maximum secondary hardware threads per processor is effective immediately and does not persist across IPL.

Additional information is available about these new APIs:

• QWCRTVPR – Retrieve Processor Multitasking Information
• QWCCHGPR – Change Processor Multitasking Information

EB33 Dynamic Platform Optimizer

Among the many features available for the new POWER7+ 770 and 780 and the POWER7 795 servers is the EB33 Dynamic Platform Optimizer. This new option for managing memory and processor resources is available for both the Power 770 Model MMD and the Power 780 Model MHD server at this time, and is available for both IBM i 7.1 and 6.1 with 6.1.1 machine code. For more information about this feature, see the RFAs at the following links:

• IBM Power 770 Enterprise Server Model 9117-MMD
• IBM Power 780 Enterprise Server Model 9179-MHD and new Power 795 model FHB features

Large Tape Library Support

IBM i 7.1 TR 5 includes support for up to 15000 storage elements in an attached 3584 or ProtecTIER library. Use Access to PTF Cover Letter to search for the cover letter for PTF MF55409 for a discussion of restrictions.

May 2012 - IBM i 7.1 Technology Refresh 4

Performance Improvement for Record Locks

Record locking algorithms have been changed to reduce the impact of large commit cycles. These changes improve the overall performance of the job obtaining these locks with minimal impact to the other jobs in the partition.

October 2011 - IBM i 7.1 Technology Refresh 3 and IBM i 6.1

Ethernet Link Aggregation

Users need their partitions to have reliable, fast connections to the network. With IBM i 7.1, a line description can be associated with up to eight Ethernet links, which are bound together into an aggregate. Outgoing traffic is spread across all of the links according to a user-selected policy, and incoming traffic by a policy configured at the switch. For example, two aggregated 1 Gbps links can carry up to 2 Gbps for the same IP interface without any additional configuration. In case a link goes down, the other links handle all traffic that would have gone on that link. Therefore, aggregation provides additional reliability while neither requiring any inactive resources in reserve nor any additional IP configuration.

The user creates a line description with resource name set to a new special value *AGG, which adds new aggregation-specific parameters. The user must choose which aggregation method to use, which policy to use for spreading outgoing frames across the aggregated links, and a list of the resources that should be aggregated. The user must select a single speed for all of them and they must all run at full duplex. Then, the corresponding ports on the switch must be bound together into an aggregate according to the switch's configuration manual. The configuration must indicate that all of the ports are always aggregated (as opposed to being negotiated according to some protocol).

When the line description is varied on, all of the links are started. Then the line description can be used by an IP interface, for example. The status of each aggregated link can be seen in a new screen under the Display Line Description command. Collection Services report data for each port rather than line in the QAPMETH file.

Ethernet Link Aggregation support requires from two to eight Ethernet ports, 10 Gbps or 1Gbps-capable, and an Ethernet switch that is capable of aggregating multiple ports (sometimes called trunking or teaming).

For more information about configuring Ethernet resources and link aggregation, refer to the IBM Documentation Ethernet Link aggregation topic.

Thin Provisioning for DS8000, and for VIOS Shared Storage Pools

With IBM i 7.1, Thin Provisioning allows configurations to be set up to use a smaller amount of real disk storage initially, and then allows for a later increase in the amount of disk storage without changing the partition's view of the storage LUN. Prior to this enhancement, the full amount of configured storage would be used at LUN initialization time.

Thin provisioning enhancement for DS8000 storage servers requires a DS8700 or DS8800 Release 6.2, available from IBM via FC #1723, or via bundles:

• IBM System Storage DS8700 - level 7.6.2.xx.xx (bundle version 76.20.xxx.xx), or later
• IBM System Storage DS8800 - level 7.6.2.xx.xx (bundle version 86.20.xxx.xx), or later

Bus level statistics for 12x loops

Enhancements have been made to Collection Services to collect and display bus level statistics for 12x loops and PCI slots. This data is stored in the Management Collection Object (*MGTCOL) created by collection services. These statistics help to give you a view of the traffic and utilization for your 12x loops and CEC PCI buses. For POWER6 and POWER7 systems, you have access to utilization statistics for the 12x loops that can provide you a view of the traffic. For POWER7 system units that have gen2 slots, in addition to the loop data, you also have access to utilization statistics for the CEC PCI buses.

Besides providing support to collect the data, IBM i 7.1 allows the data to be exported to the new QAPMBUSINT file, and places additional PCI bus data for hubs into the QAPMBUS file. Software prerequisite: IBM i 7.1 TR 3.

For IBM i 6.1, only collection support is provided, but that collection object can be moved to an IBM i 7.1 partition for displaying. To access this data you must restore the *MGTCOL to an IBM i 7.1 system and manually run CRTPFRDTA against it. The data is exported to files as described for the 7.1 support. Software prerequisite: PTF SI43498.

May 2011 - IBM i 7.1 Technology Refresh 2

More advanced multipath algorithms

For IBM i clients who want to move their performance-sensitive workloads to external or VIOS-attached storage, advanced multipath algorithms are now available that can improve performance in certain multipath configurations. The new algorithms in IBM i 7.1 allow the system to be more tolerant of paths with different performance characteristics. They automatically attempt to choose the path that has the most capacity for the I/O, thus using higher-performance paths. This is particularly useful in environments where the adapter is being shared among multiple partitions and all adapters do not have symmetric workloads.

Enablement for IBM Disk Sanitizer PRPQ to include SSDs

The capability of PRPQ 5799SD1 IBM Disk Sanitizer for i has been extended to allow Solid State Drive (SSD) units to be sanitized using the SSD's native secure erase commands. These commands erase both the user addressable and the over-provisioned areas of the SSD. The capability is invoked by issuing the ssdsanitize command from the macro interface from either the Dedicated Service Tools (DST) menu or the System Service Tools (SST) menu. IBM i 7.1 Technology Refresh 2 includes enablement for this function.

For additional information, see the Disk Sanitizer PRPQ.

October 2010 - IBM i 7.1 Technology Refresh 1 and IBM i 6.1

Installation of new partitions via network

IBM i can now support the install of new partitions from images on the network. An IBM POWER7 processor-based server can be installed without using physical media but instead uses IBM i 7.1 or IBM i 6.1 optical images that are stored on a network file server. For IBM i 7.1, refer to APAR II14482 for PTF and setup information. For IBM i 6.1, refer to APAR II14310 for PTF and setup information.

CEC node level mirroring

Storage availability has been enhanced. A system that has multiple CEC nodes now mirrors disk units in a way that allows it to survive a node outage and that allows a concurrent maintenance repair of a CEC node. When starting mirroring, the operating system takes into consideration the CEC node under which the disks are located, and attempts to place the two subunits of a mirror protected pair under different CEC nodes. This allows concurrent maintenance of a CEC node since the two subunits of each mirrored disk unit pair are located under a different CEC node, thereby allowing at least one subunit of each mirrored disk unit pair to remain operational during the maintenance operation. After installing the PTF Group that contains this function, the user may want to consider ending and restarting mirroring to recalculate the mirror protected pairs. There is an Advanced Analysis macro called "levelofprotection", accessible through SST or DST, that allows a user to verify the level of protection for each mirrored pair.