IBM BS029ML - WebSphere Portal Server Self Help Manual page 64

Figure 2-8 gives an overview of the DB2 HADR.
DB01
pSeries 570
4x 1.65 Ghz
AIX 5.3.0.0 + HACMP 5.3
8 Gb RAM
DB2 UDB ESE
V8.2 FP 10
(primary)
Tape Library
Figure 2-8 DB2 HADR
Without HADR, the length of time it takes to cut over from a database failure is unpredictable.
It can take several minutes or hours before the failure is solved and the database is available.
HADR enables failover and fallback between the two systems. The Standby database can
take over as the Primary database with full DB2 functionality. After the failed old Primary is
repaired, it can rejoin the HADR pair as a Standby database if the two copies of the database
can be made consistent. After the original Primary database is reintegrated into the HADR
pair as the Standby database, a failback operation can be performed so that the original
Primary database is once again the Primary database. HADR requires the same hardware,
OS, and DB2 software on the two systems (except for some minor differences).
A high availability mechanism is still a requirement, as DB2 HADR does not have a fault
tolerant detection feature. In addition, if the Log transfer network is down, HADR takeover
cannot be done. So this network is very important. In this configuration, a dedicated Gigabit
Ethernet segment is used in conjunction with Network Interface Backup (NIB) for redundancy.
Note that an outage at the Log transfer network would cause the Primary to drop out of
communication with the Standby (and if in Peer state, would cause the Primary to drop out of
Peer state and run independent of the Standby). Once the network is repaired, the Primary
and Standby would be able to eventually come back into Peer. Thus, a network outage at the
Log transfer network would not result in a failure to process transactions, as seen by clients.
An alternative to HACMP is Tivoli System Automation (TSA). TSA is now bundled with DB2
for AIX (as of DB2 ESE 8.2 FP 13) in the same manner, and same licensing terms, as Linux
(TSA bundled with DB2 ESE 8.2 on Linux). All HACMP or TSA has to do is detect a node
50
IBM WebSphere Portal V6 Self Help Guide
Ethernet (Service Network)
en0
en2
en1
en4
en3
Gigabit Ethernet
(for HADR Log transfer)
tty1
RS232C (for HACMP)
SAN Storage
Area for Node 1
Data
Logs
DB02
pSeries 550
2x 1.65 Ghz
AIX 5.3.0.0 + HACMP 5.3
4 Gb RAM
en0
DB2 UDB ESE
en2
V8.2 FP 10
(standby)
en1
en4
en3
tty1
SAN Storage
Area for Node 2
Data
Logs
Tape Library