z/OS MVS Setting Up a Sysplex
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Considerations for all couple data sets

z/OS MVS Setting Up a Sysplex
SA23-1399-00

The following considerations apply to sysplex couple data sets as well as to couple data sets that hold policy data.
  • A Couple Data Set Can Be Formatted with the Couple Data Set Format Utility

    IBM® provides a couple data set format utility, IXCL1DSU, to format the couple data sets. Sample JCL to invoke the format utility, along with explanations of the information you need to specify when formatting each type of couple data set is provided in Format utility for couple data sets.

    (The WLM couple data set can also be formatted interactively using the WLM administrative application. See z/OS MVS Planning: Workload Management.)

  • Format the couple data set to support the number of systems in the sysplex

    All systems in a sysplex that use Start of change function-specific data must be able to access both the primary and alternate couple data sets that were created to contain that data. The value specified on the MAXSYSTEM parameter for function couple data sets should match the value specified for MAXSYSTEM when formatting the sysplex couple data set. End of change

  • An Alternate Couple Data Set Is Highly Recommended
    To avoid a single point of failure in the sysplex, IBM recommends that, for all couple data sets, you create an alternate couple data set on a different device, control unit, and channel from the primary.
    Note: For storage subsystem devices like the 2105 ESS, which emulates multiple logical control units, you should attempt to place the alternate couple data set on a different physical control unit if possible.

    Information about the sysplex and the services that use couple data sets is maintained in both the primary and alternate couple data sets concurrently. If the primary data set fails, the sysplex automatically makes the alternate the primary.

    Besides preventing a single point of failure, having an alternate couple data set has other advantages. You can issue the SETXCF COUPLE,PSWITCH command to dynamically make the alternate couple data set the primary. Thus, in a non-disruptive way, you can:
    • Expand the size of the primary couple data set by switching to an alternate that is formatted to accommodate additional ITEMS appropriate to that type of couple data set.
    • Change the couple data set to a different device, by defining the alternate couple data set on a different device from the primary. Thus, you can circumvent device errors or take advantage of the characteristics of a different device.
    Then, take a formatted spare couple data set (or format a new alternate couple data set) and define it to XCF:
    SETXCF COUPLE,ACOUPLE=(alt-dsname[,alt-volume]),TYPE=dstype

    The system issues message IXC267E when the primary or alternate couple data set of any type is removed, to indicate that the sysplex is operating without an alternate. Since the remaining couple data set constitutes a single point of failure, consider defining automation rules to respond to message IXC267E and automatically activate a pre-formatted spare couple data set using the SETXCF COUPLE,ACOUPLE command.

  • An Alternate Couple Data Set Must Be At Least As Large As the Primary

    After a primary couple data set has become active in the sysplex, the couple data set can be removed from service by adding an alternate couple data set (if one does not currently exist) to the sysplex and by issuing a “PSWITCH” to make the alternate couple data set the primary. When an alternate couple data set is added to the sysplex, it must be at least as large as the current primary data set. That is, the alternate couple data set must be formatted with ITEM values equal to or greater than the ITEM values that were used to format the primary couple data set. For example, if the primary sysplex couple data set was formatted for 50 XCF groups and 200 XCF group members, the new sysplex couple data set must be formatted for 50 or more XCF groups and 200 or more XCF group members. Likewise, the values specified for the MAXSYSTEM parameter and if applicable, the MAXGROUP/MAXMEMBER parameters, must be equal to or greater that the values that were used to format the primary couple data set.

    When formatting couple data sets consider specifying parameter values commensurate with your hardware and software configuration. You might wish to add some contingency (for example, 20%) to individual parameter values to accomodate growth.

  • Avoid Overspecifying Parameter Values

    Do not excessively overspecify parameter values when formatting couple data sets. Formatting couple data sets with parameters that are overspecified results in wasted space and degraded couple data set performance. Furthermore, if the current primary couple data set is formatted with parameter values that are overspecified and it becomes necessary to format and switch to a couple data set with smaller parameter values, the sysplex might need to be reinitialized in order to switch to the smaller format couple data set.

    It is legitimate to add some contingency to parameter values to allow for growth. However, do not add excessive contingency. Remember that you can always increase couple data set capacity by issuing “PSWITCH” to switch to an alternate couple data set that is formatted with parameter values greater than the parameter values that were used to format the primary couple data set.

  • A Spare Couple Data Set Is Recommended

    When an alternate couple data set replaces a primary, the original primary data set is deallocated, and there is no longer an alternate couple data set.

    Because it is advisable to have an alternate couple data set always available to be switched, for each couple data set you plan to use, consider formatting three data sets before the IPL. For example, for the sysplex couple data set, you might format the following:
    SYS1.XCF.CDS01 - specified as the primary couple data set
    SYS1.XCF.CDS02 - specified as the alternate couple data set
    SYS1.XCF.CDS03 - a spare

    Then, if the alternate (CDS02) becomes the primary, you can issue the SETXCF COUPLE,ACOUPLE command to make the spare data set (CDS03) the alternate.

  • The Couple Data Set Must Not Exist Prior to Formatting

    The format utility cannot use an existing data set. This prevents the accidental re-formatting of an active couple data set. You must delete an existing couple data set before reformatting it.

  • The Couple Data Set Must Be Formatted, Not Copied

    The contents of an existing, in-use couple data set should not be copied to another data set for use by another sysplex, or another instance of a sysplex. If a couple data set must be moved to another volume, the preferred way to do this would be to use the IXCL1DSU couple data set format utility to format a new, empty couple data set of the required type on the required volume (making sure that the formatting parameters for the data set provide enough capacity to bring the couple data set into use with the currently-active primary couple data set of that type), and then bring the data set into use as the alternate couple data set using the SETXCF COUPLE,ACOUPLE command. Then, if you want, the new data set might be made the active primary couple data set using the SETXCF COUPLE,PSWITCH command.

    Note that couple data sets contain the following:
    • The sysplex name of the sysplex to which they pertain and in which they will be used.
    • Sysplex ownership records that contain information about the instance of the sysplex that last used the couple data set.
    • Various real-time status and policy information that pertains to the function for which the couple data set is formatted, for example, SYSPLEX, CFRM, SFM, and so forth.
    Simply making a copy of a couple data set creates a snapshot of all of this information at a point in time; as time then passes, this snapshot of the function's usage and policy information can become more and more out-of-date. Subsequently bringing this couple data set copy into use as a primary couple data set in a sysplex (for example, by initializing a sysplex specifying the couple data set copy as a PCOUPLE) can cause serious problems, as the status information in the couple data set copy is an obsolete snapshot rather than the most up-to-date sysplex information, and might experience inconsistencies with more current information stored in databases or other types of data sets that the sysplex is using.

    When re-initializing a sysplex, the most current couple data sets previously in use in the sysplex should always be used. Regressing to use older copies of couple data sets, or couple data sets from another sysplex instance, is not recommended.

    For additional information and recommendations regarding the use of remote copy technologies (for disaster recovery) to replicate couple data sets, see Planning disaster recovery actions.

  • A Multiple Extent Couple Data Set Is Not Supported

    For the sysplex couple data set, the format utility determines the size of the data set based on the number of groups, members, and systems specified or defaulted, and allocates space on the volume specified for the data set. There must be enough contiguous space available on the volume for the sysplex couple data set.

    For the couple data sets that hold policy information, the format utility determines the size of the data set based on parameter values within the policy.

  • A Couple Data Set Cannot Span Volumes

    XCF does not support multiple-volume data sets.

  • A Couple Data Set Can Be Used by Only One Sysplex

    The name of the sysplex for which a data set is intended must be specified when the data set is formatted. The data set can be used only by systems running in the sysplex whose name matches that in the couple data set.

  • A Couple Data Set Can Share a Volume with Other Data Sets
    However, if you decide to format a couple data set on a volume with other data sets,
    • Avoid a volume that has the RESERVE macro issued against it.
    • Avoid a volume with high use data sets, such as page or spool data sets.
  • Performance and Availability Considerations
    The placement of couple data sets can improve performance, as well as availability. For maximum performance and availability, each couple data set should be on its own volume. While this is an expensive approach, placing some data sets on different volumes from others makes sense.
    • The primary sysplex couple data set should be on a different volume from the primary CFRM couple data set.
    • All other primary couple data sets can reside on one volume, and all other alternate couple data sets can reside on another volume, as shown in the following table. However, be sure to monitor these data sets, and consider placing any high-activity data set on its own volume. (For example, the LOGR couple data set might be a candidate for its own volume if it experiences sufficient I/O activity to warrant its placement on a volume separate from the sysplex or CFRM primary couple data set.)
      Volume X Couple Data Sets      Volume Y Couple Data Sets
      -------------------------      -------------------------
        Sysplex (Primary)              Sysplex (Alternate)
        CFRM (Alternate)               CFRM (Primary)
        SFM (Primary)                  SFM (Alternate)
        WLM (Primary)                  WLM (Alternate)
        ARM (Primary)                  ARM (Alternate)
        LOGR (Primary)                 LOGR (Alternate)
        ...and so forth                ...and so forth
    • Place couple data sets on volumes that are attached to cached control units with the DASD fast write (DFW) feature. Although this recommendation applies to all couple data sets in any size sysplex, it is more critical the more systems you have in the sysplex. Those couple data sets most affected are the sysplex couple data set and the CFRM couple data set.
    • Place couple data sets on volumes that are not subject to reserve/release contention or significant I/O contention from sources not related to couple data sets. This is true even if the I/O contention is sporadic.
    • Do not excessively overspecify parameter values when formatting couple data sets.
    • Some sysplex monitoring tools may generate a large amount of I/O to the CFRM couple data set, which may affect system performance. When using this type of tool, be aware of the performance implications. If you are designing a sysplex monitoring tool, see z/OS MVS Programming: Sysplex Services Reference and z/OS MVS Programming: Sysplex Services Guide for additional information about how XCF and XES macro services interact with the CFRM active policy.
    If a system cannot access a couple data set for an extended period of time (for example, if the volume on which it resides is reserved by another system), MVS™ switches to its alternate couple data set. To minimize sysplex disruption, when you use DFDSS (or another data mover) to back up a volume with a couple data set, it is recommended that you:

    The system drives periodic I/O to all couple data sets, which acts to verify their continued accessibility. In the case of the sysplex couple data sets, the normal system status updates serve this purpose. For other couple data sets, which are not otherwise frequently accessed, the system "probes" about once per minute to determine whether the data set can still be accessed. Any data set found to be inaccessible due to I/O or device error is removed from service in accordance with XCF's normal recovery protocols. This ensures that problems are surfaced in a timely manner, preventing an undetected single point of failure and allowing the installation to provide an accessible replacement.

  • Security Considerations

    It is the responsibility of the installation to provide the security environment for the couple data sets. Consider protecting the couple data sets with with the same level of security as the XCF address space (XCFAS).

    z/OS Security Server RACF System Programmer's Guide explains how to add a started procedure to the started procedures table.

  • The ADRDSSU Utility Can Dump the Contents of a Couple Data Set

    If you experience problems with a couple data set or with the usage of a couple data set, you can use ADRDSSU to dump the data set. For example, if you are having problems with coupling facility structures, use ADRDSSU to dump the CFRM data set.

    The following JCL could be used to dump a couple data set named SYS1.PRIMARY, which resides on the 3380 volume, SHR001.
    //DUMP  JOB MSGLEVEL=(1,1)
    //STEP1     EXEC PGM=ADRDSSU,REGION=4M
    //SYSPRINT  DD SYSOUT=*
    //DD1       DD DISP=SHR,VOL=SER=SHR001,UNIT=3380
    //SYSIN     DD *
         PRINT  DATASET(SYS1.PRIMARY) INDDNAME(DD1)
    /*
  • A Couple Data Set Can Be Defined on an SMS-Managed Volume
    You can define a couple data set on a DASD volume that the Storage Management Subsystem (SMS) manages. When placing a couple data set on an SMS-managed volume, consider the following:
    • Ensure that all systems that need to can access the couple data set. Catalog both the primary and alternate couple data sets. If the catalog is not shared by all systems, ensure that the couple data set is cataloged on each system. IBM recommends that you catalog the couple data set in the master catalog, rather than a user catalog. If you catalog the couple data set in a user catalog, the volsers must be specified in COUPLExx in order for the data sets to be found during IPL, so there is no benefit in using a user catalog.
    • To prevent a single point of failure, define the primary and alternate couple data sets to different control units and different volumes.

      The couple data sets can reside on SMS-managed and non-SMS-managed volumes. For example, the primary couple data set can reside on an SMS-managed volume while the alternate can reside on a non-SMS-managed volume.

    • Using the Interactive Storage Management Facility (ISMF) panels, define a storage class for the SMS-managed volume with GUARANTEED SPACE=YES. When you use the format utility to format the primary and alternate couple data sets, you can specify the appropriate storage classes (STORCLAS).

      You must ensure that the volumes you choose for the couple data set are available to all systems that require it. Specifying the guaranteed space attribute for the storage class ensures that SMS can select from these eligible volumes.

    Protect the couple data set from being migrated or deleted by DFHSM. To prevent accidental expiration or migration of the volume by DFHSM, define the following management class attributes through the ISMF panels:
    • EXPIRE AFTER NON-USAGE=NOLIMIT
    • EXPIRE AFTER DATE/DAYS=NOLIMIT
    • COMMAND or AUTO MIGRATE=NONE

    You can use the format utility to format the couple data set on an SMS-managed volume by specifying the STORCLAS, MGMTCLAS, or VOLSER keywords. For a description of these keywords, see Format utility for couple data sets.

    For information about SMS storage classes and management classes, see Storage Management Subsystem Migration Planning Guide.

  • Start of changeUse of Remote Copy Technology with Couple Data Sets

    See Planning disaster recovery actions for considerations on the use of remote copy technology with couple data sets.

    End of change

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014