曝光台 注意防骗
网曝天猫店富美金盛家居专营店坑蒙拐骗欺诈消费者
Network Best Practices [2].
Observer
For Disaster Recovery requirements, install the Observer in a location separate
from the primary and standby data centers. This prevents any single event from
impacting 2 out of 3 of the members of a Fast-Start Failover configuration.
Installing the Observer is simple – all that is required is the Oracle Administrator
Client installation. The Observer is also included with the Oracle Database
Enterprise or Oracle Personal Edition. An Oracle instance is not required on the
Observer system.
Fast-Start Failover Threshold
A Fast-Start Failover occurs when the Observer and the Standby Database both
lose contact with the primary database for a period of time that exceeds the Fast-
Start Failover Threshold. The correct setting weighs the trade-off between the
fastest possible failover (thus minimizing downtime), and triggering unnecessary
failovers due to fleeting network irregularities. Recommended settings for the
FastStartFailoverThreshold are:
• Single instance primary, low latency, reliable network = 10-15 seconds
• Single instance primary, high latency network over WAN = 30-45 seconds
• RAC primary = (RAC miscount + reconfiguration time) + 24-40 seconds
Impact of Network Tuning
937
10.8
0 200 400 600 800 1000
Tuned
Default
Mbits/sec
Figure 3 – Impact of Network Tuning
Tuning TCP Send/Receive Buffers, Device Network Que Size & SDU
Fast-Start Failover – Oracle Data Guard 10g Release 2 Page 14
Monitoring
Because a Fast-Start Failover will not occur unless the primary and standby are in a
synchronized status (insuring zero data loss), it is important to respond quickly to
any event such as a network outage or standby server crash, so that Data Guard
can quickly resolve any resulting redo gap and return the standby configuration to
a synchronized status.
Enterprise Manager can be used to continually monitor configuration status and
automatically notify administrators of events that require attention. Alternatively
the state of the configuration can be monitored via the FS_FAILOVER_STATUS
column of the V$DATABASE view. A SYNCHRONIZED status means that the
primary and standby are in sync and a Fast-Start Failover is possible.
UNSYNCHRONIZED status means the standby has not received all of the redo
generated by the primary database and a Fast-Start Failover cannot occur.
Because a Fast-Start Failover requires two of the three members in the
configuration to agree, Observer status should also be monitored, either through
the Enterprise Manager GUI, or by monitoring the Observer via the
FS_FAILOVER_OBSERVER_PRESENT column of the V$DATABASE view.
Flashback Database Configuration
Configure Flashback Database on both the primary and the standby databases.
Flashback Database is used to automatically reinstate a failed primary as a standby
database to the new primary in the event that a failover occurs. This can occur as
long as the old primary can be restarted and the failover operation is a zero data
loss failover executed by Fast-Start Failover.
Oracle recommends that when used in a Fast-Start Failover context,
DB_FLASHBACK_RETENTION_TARGET is increased from its default to a
setting of 10 minutes. Even this low retention target is a conservative value when
used by Fast-Start Failover. Note that if Flashback Database serves the additional
function of protection against user error and corruption, then an extended
flashback retention period should be set for an amount of time required to achieve
these goals.
Configurations with Multiple Standbys
It is not unusual for a Data Guard configuration to include more than one standby
database for a single primary database. Often a “local” standby is utilized to
achieve zero data loss protection by utilizing a low latency LAN capable of
supporting synchronous data protection without incurring the performance
overhead of a high latency WAN. The second standby is always geographically
remote; 100’s to 1,000’s of miles away from the primary production site. The
second standby always utilizes asynchronous data protection to avoid the
performance hit of a high latency WAN. Using Fast-Start Failover in such a
configuration, the local standby would be designated as the failover target and
would be the system, along with the primary production database, that the
Fast-Start Failover – Oracle Data Guard 10g Release 2 Page 15
Observer would monitor. At failover, Fast Start Failover would transition the local
standby into the primary role, and the remote standby would transition seamlessly
to the role of standby database to the new primary – affording continuous data and
中国航空网 www.aero.cn
航空翻译 www.aviation.cn
本文链接地址：
航空资料22(58)
