A HBA RAC design will have many physical network
connections in order to provided true
redundancy.
Note that all of the physical connections
are not shown in diagram 4.2.
In an implementation, each component is
physically connected usually via a network
interface card (NIC) or host bus adapter (HBA)
interface.
NIC or HBA interfaces should be the fastest
possible, especially in the case of the cluster
interconnect and disk connect.
For example, using Gigabit Ethernet
for the interconnect seems very fast when
compared to older network technologies. However,
a single Gigabit Ethernet NIC should not be
considered more than adequate in all cases.
The Oracle RAC design may require NIC
teaming/bonding. Upgrading to 10 Gigabit
Ethernet is another option to increase
bandwidth.
10 Gigabit Ethernet speed is 1,250 MB/s
versus Gigabit Ethernet’s 125 MB/s.
Failed NIC interfaces result in the loss of that
component unless a second NIC card is failed
over to immediately. A failure of a
non-redundant HBA results in loss of connection
to the disk array.
Redundancy is the key.
Not having multiple NICs and HBAs on a
mission critical Oracle RAC is a poor design.
Failure of the InterConnect Switch / Memory
The interconnect switch is a core part of the
RAC Architecture.
The switch should be an enterprise grade
managed switch.
10 Gigabit Ethernet
switches are not cheap, but could increase
performance significantly.
Various server vendors provide advanced memory
fault protection features such as Error
Correcting Code (ECC), online spares, and
mirroring.
Provide Redundancy at Each Level
It is easy to see that redundancy at the
hardware level is vital. At each level of the
hardware layout an alternate access path must be
available. Duplicating all equipment and
configuring the automatic failover capabilities
of the hardware reduce the chances of hardware
failure to virtually nil.
By providing the required levels of redundancy,
the system becomes highly available. Once there
is an HA configuration, it is up to the manager
to plan any software or application upgrades to
further reduce application downtime. In Oracle
Database 11g, rolling upgrades are supported,
further increasing reliability.
It is highly recommended that all
upgrades be tested on non-production systems
first.
Oracle 11g provides a new testing tool
called Database Replay.
More information about the Enterprise
Edition Database Replay capabilities can be
found at oracle.com.
At the SAN level, appropriate duplication
software should be used to ensure the SAN arrays
are kept synchronous. Oracle Database 11g allows
for use of Oracle Automatic Storage Management
or ASM.
ASM can provide striping to improve
availability.
DBA and User Error Protection
Oracle provides many data protection utilities
and features that should be tested and
understood so that an emergency does not involve
a learning curve.
RMAN, of course, is Oracle’s backup
utility.
Flashback is new as of Oracle 10g.
Here is a list of the various components
of the Flashback package:
-
Flashback Query
-
Flashback Version Query
-
Flashback Transaction Query
-
DBMS_FLASHBACK package
-
Flashback Transaction
-
Flashback Data Archive (Oracle Total Recall)
-
Flashback Table
-
Flashback Drop (Recycle Bin)
-
Flashback Database
