HyperPAV

Parallel Access Volumes (PAV) is the concept of using multiple devices or aliases to address a single ECKD disk device.

If there is no aliasing of disk devices then only one I/O transfer can be in progress to a device at a time, regardless of the actual capability of the storage server to handle concurrent access to devices. Parallel access volume exists for Linux on System z in the form of PAV and HyperPAV. PAV and HyperPAV are optional features that are available on the DS8000 series.

HyperPAV has been supported on Linux on System z since Red Hat Enterprise Linux 6.0, SUSE Linux Enterprise 10 SP4, SUSE Linux Enterprise 11, Red Hat Enterprise Linux 5.9. This study uses HyperPAV only because it is much easier to administer and provides the greater flexibility.

The Base and HyperPAV devices are defined on the storage server and on the IOCDS on System z, where the DASD devices are defined with unit addresses. A LCU is a logical set of DASD devices on a DS8000 series disk storage unit and it can have up to 256 possible DASD device addresses from 00 to ff. After the base devices are defined any remaining device numbers in an LCU can be used as an alias by the system to access any of the base addresses in the same LCU.

While the usage of PAV devices requires the usage of the multipath daemon and its configuration, the advantage of HyperPAV is that it only needs to be enabled on Linux, everything else is handled by the kernel, and is transparent to the user.

HyperPAV devices on a SUSE Linux Enterprise distribution can be activated by using yast or the tool dasd_config, for example, dasd_configure 0.0.c062 1

This also creates the required udev rules.

Note:
  • Ensure that the devices are not excluded from the device list via cio_ignore in zipl.conf
  • There is at no time a fixed connection between a HyperPAV device and a Base device!