HP XP P9500 Storage - Manual (Page 37)

• Reserved by mainframe host. When a Data Exchange volume is reserved by the mainframe host,
Data Exchange operations cannot be performed on that volume because the Data Exchange access
from the open-system host will terminate unsuccessfully. Open-system access other than read or
write I/Os can be executed successfully.
NOTE:
Open-system access to a mainframe-reserved volume may complete successfully if the open-system
retries the operation after the reserve is released. However, since the time interval before a retry
varies depending on the open-system platform and the mainframe application that issued the
reserve, the success of retry operations on reserved volumes cannot be guaranteed.
• Reserved by open-system host. When a Data Exchange volume is reserved by the open-system
host, Data Exchange operations can be performed only from the host that reserved the volume.
Data Exchange operations from any other open-system host will terminate unsuccessfully. Open-
system reserve does not affect mainframe access to the Data Exchange volume.
• Unreserved. When a Data Exchange volume is not reserved by any mainframe or open-system
host, Data Exchange operations can be performed from any open-system host using Data Exchange.
All mainframe hosts and all open-system hosts have access to unreserved volumes.
The user should implement exclusive access control and job coordination at the system level for the
Data Exchange volumes. The user should also take the following steps to avoid I/O contention problems
for the Data Exchange volumes:
• Open-system access. When the open-system host needs to access a Data Exchange volume, vary
the volume and its channel path offline from all mainframe hosts.
• Mainframe access. When the mainframe host needs to access a Data Exchange volume, stop all
open-system access to the corresponding LU. For AIX, vary off the volume groups. For Windows,
use unaccess. Do not use any open-system program that accesses unmounted LUs (for example,
AIX SMIT and HP-UX SAM).
Bidirectional Data Transfer
Data Exchange supports bidirectional data transfer for both fixed-length and variable-length mainframe
datasets. Bidirectional data transfer involves transferring data from mainframe datasets to open-system
files and then back to the original mainframe datasets again. The requirements for bidirectional data
transfer are:
• For all FXmto operations, do not specify the RDW option. If the RDW option is specified for an
FXmto data transfer, the subsequent FXotm target dataset will not be compatible with the original
dataset.
• For FXmto with fixed-length datasets, do not specify the delimiter option, since the data entities
are extracted by length. If you add delimiters for the FXmto transfer, the subsequent FXotm target
dataset will not be compatible with the original dataset.
• For FXmto with variable-length datasets, you must add delimiters but not padding. If delimiters are
not added or if padding is added for the FXmto transfer, the subsequent FXotm target dataset will
not be compatible with the original dataset.
• For FXotm operations, do not specify the delimiter option if the source file contains the same
characters as the delimiter (CR or LF) but are being used for purposes other than delimiting data
entities. If you specify the delimiter option for FXotm, FCU will interpret all occurrences of the
specified delimiter characters as delimiters, which can create a dataset with corrupt records or
generate an error condition.
HP StorageWorks P9000 Data Exchange User Guide 37