HP StorageWorks XP Remote Web Console Software - Manual (Page 34)

34 Hitachi TrueCopy z/OS for HP XP12000/XP10000 and SVS 200 storage systems
Timer Type. This group option determines how the MCU will acquire the time-stamp for the TC390A
recordsets:
• System. When the System timer option is selected, the MCU acquires the time-stamp information for
each recordset as follows. When a TC390A pair is established, the MCU reports state-change-interrupt
(SCI) to all hosts. The host then issues a series of sense group commands to determine the device status
change and the MCU returns the same response as if the device had been added to an XRC session to
activate I/O time-stamping for the device. After I/O time-stamping is activated, the MVS IOS routine
attaches the time-stamp information (contents of time-of-day (TOD) clock) to each write I/O operation
for the device. The time-stamp indicates the time that the update was generated during start subchannel
(SSCH) at the main host system and the time-stamp is transferred to the MCU at the beginning of each
I/O operation.
• Local. The Local timer option enables the MCU to generate the time-stamp for each update I/O using
its own internal clock, rather than using the SMS I/O time-stamp.
• None. The None timer option should only be selected when establishing TC390A pairs in the reverse
direction (from secondary to primary). When the None option is selected, the MCU still acquires the
time-stamp information from the host I/O time-stamping function.
Time Out [Copy Pending]. This group option specifies the maximum delay allowed for TC390A copy
operations. For instructions on selecting the correct Time Out [Copy Pending] setting for your operational
environment, see ” Adding consistency groups (Add CT Group) ” on page 105.
The RCU will suspend all R-VOLs in the group when:
• The RCU has not had any communication from the MCU within the specified time. This situation could
indicate a disaster or failure at the primary site.
• The RCU has been receiving recordsets from the MCU(s) but has not been able to settle a recordset
within the specified time. This situation may indicate that the RCU does not have enough resources to
handle the remote copy and I/O workloads.
RCU Ready. This option specifies the maximum delay for re-establishing MCU-RCU communications
following MCU power-off. During MCU power-on, the MCU re-establishes communication with all
registered RCUs. If it is not able to re-establish communication with an RCU within the specified time, the
MCU suspends all affected TC390A volume pairs.
Group consistency time
During normal TC390A operations, the consistency time (C/T) of a group corresponds to the time-stamp
value of the most recently settled recordset at the RCU. The consistency time for the group is indicated as
part of the TC390A R-VOL pair status (also displayed by the CQUERY TSO command to the R-VOL). As the
main system continues to update the TC390A M-VOLs, the difference between the current system time and
the group consistency time indicates the amount of time that the R-VOLs are behind the M-VOLs. The M-VOL
updates that take place during this time may be lost when a disaster occurs.
When a TC390A volume pair is suspended, the C/T of the suspended R-VOL is frozen. If the RCU can
ensure the update sequence consistency between the suspended R-VOL and the other R-VOLs in the
consistency group, the R-VOL C/T is frozen at the latest consistency time of the group. Otherwise, the
R-VOL C/T is frozen at the time-stamp value of the most recent update that was successfully copied to the
R-VOL. The C/T of a suspended R-VOL may be older than the C/T of other R-VOLs in the group, and if the
entire group was not also suspended, the consistency time of the group is still ticking. For suspended
TC390A R-VOLs, the TC390 Pair Status window displays whether the C/T was frozen to the group or
R-VOL time.
After you have established TC390A operations, you should monitor the consistency time of each group at
the RCU(s). If the average delay is longer than your disaster recovery design can accept, you should
consider adding remote copy resources (for example, paths or cache) and/or reducing the I/O workload
to improve disk array performance. If the delay between the M-VOL update and the corresponding R-VOL
update reaches the time specified by the Time Out [Copy Pending] group option, the MCU will suspend all
affected volume pair(s) due to the heavy I/O workload (at MCU or RCU). To prevent timeout errors, you
can increase the Time Out [Copy Pending] value, reduce I/O workload, and/or add remote copy
resources.