Policy, Horcm.conf, Policy horcm.conf – HP XP7 Storage User Manual
Page 176
on the open system if RAID Manager is capable of delivering the timestamps and committing the
Journal data with a CTQ-Marker.
Thus, RAID Manager supports Cnt Ac-J MxN for open systems in the four ways described in
Policy
All data consistency of the CT group across multiple storage systems is maintained by RAID
Manager. The storage system supports only basic functions; there is no relation between storage
systems in order to simplify testing and configurations.
RAID Manager supports Continuous Access Journal MxN in the following ways.
(1) Delivering the timestamp (CTQ-Marker)
RAID Manager (HORCM daemon process) makes a table for the groups registered to the horcm.conf
as HORCM_CTQM with startup, and makes the threads for each group that delivers the same
timestamp with an increment to the multiple storage systems configured in a group. The thread for
a group delivers the same timestamp with increments, as far as a group configured Continuous
Access Journal in the PAIR status.
The timestamp is delivered by using Freeze/Q-Marker & Run way as default. The timestamp is
maintained in each storage system. RAID Manager includes this timestamp with startup, and then
delivers the same timestamp with increments to each storage system.
(2) Arbitrating/committing the journal data with CTQ-Marker
The thread for a group on HORCM compares the timestamp of S-JNL on each storage system (RCU)
as far as a group configured Continuous Access Journal in PAIR state. Once HORCM detects a
matching point of the timestamp (CTQ-Marker) on all storage systems (RCU), it issues an order to
commit the Journal data with CTQ-Marker to each storage system (RCU).
(3) Propagating Error suspend
The thread for a group on HORCM delivers the same timestamp with increments as far as a group
configured Continuous Access Journal in PAIR state. If the PSUE/PFUS state detects at least one
storage system, then it notifies another storage system to suspend PSUS in order to keep the state
consistent in the CT group. Then the thread stops to deliver the timestamp, and keeps monitoring
its CT group with interval of HORCM_CTQM until it becomes PAIR state by next pair-resync.
(4) Committing the Journal data inside pairsplit command
The pairsplit command makes a suspending state on PAIR state, and inherits to compare the
timestamp of S-JNL on each storage system (RCU). If it detects a matching point of the timestamp
(CTQ-Marker) on all storage systems (RCUs), then it issues an order to commit the Journal data
with Q-Marker to each storage system (RCU), and repeats it until it detects an EOM (End Of Marker)
of CTQ-Marker with the pairsplit command.
horcm.conf
RAID Manager supports Cnt Ac-S_Sync group across multiple storage systems, but it does not allow
Continuous Access Asynchronous (Cnt Ac-J) group across multiple storage systems. Therefore, RAID
Manager needs to add the group definition (HORCM_CTQM) in order to allow making Continuous
Access Asynchronous(Cnt Ac-J) group across multiple storage systems. Then the HORCM daemon
process delivers the timestamps (called the consistency Q-Marker), and commits SVOL Journal data
with Q-Marker to the defined group.
(1) Defining to control Cnt Ac-J MxN
RAID Manager supports a way to specify consistency Q-Marker to the specified group by adding
"HORCM_CTQM" as a keyword in horcm.conf (see example below).
176
Data replication operations with RAID Manager