Connection Time Estimates During Data Guard Switchover or Failover
The settings in the connect string allow for the following maximum times to connect during switchover or failover.
-
Data Guard Switchover:
RETRY_COUNT X RETRY_DELAY = 100 x 3 sec = 300 sec.
-
Data Guard Failover:
(3 SCANs x TRANSPORT_CONNECT_TIMEOUT) + (RETRY_COUNT x (RETRY_DELAY + (3 SCANS x TRANSPORT_CONNECT_TIMEOUT))) = (3 x 1) + (100 x (3 + (3 x 1)) = 3 + 600 = 603 sec
After Data Guard switchover / Data Guard failover to clu-site2, initial connections to
clu-site2 take 3 seconds when clu-site1 is down (use of a connection pool helps mitigate
this delay). When clu-site1 is reachable again (when it becomes a standby) connections are
nearly instantaneous because the listener on the standby will answer immediately that the
service is not there, prompting the client to connect to the other
ADDRESS_LIST
.
-
If the switchover or failover completes much earlier than the maximum time, the application will experience less impact.
-
Increase RETRY_COUNT if your system might take longer than 300 sec to complete a switchover or failover. If you need more time to complete a Data Guard switchover, then change RETRY_COUNT greater than 100.
- If you aren't using Oracle Clusterware, then your HOST address will not reference a SCAN
VIP but a single VIP. This means that
TRANSPORT_CONNECT_TIMEOUT
must be set to higher or lower values to account for network latency.