I am a DBA in Japan. I'm doing TAF implementation now, but I can't understand TAF well.
We have two DB servers(cwdb1,cwdb2), two application servers(cwap1,cwap2) and clients. I use cwap2 for TAF test. The following is tnsnames.ora on cwap2.

# TNSNAMES.ORA Configuration File:/opt/oracle/product/8.1.6/network/admin/tnsnames.ora
# Generated by Oracle configuration tools.
CWDB =
(DESCRIPTION =
(FAILOVER = ON)
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = TCP)(HOST = xxx.xxx.xxx.x)(PORT = 1521))
)
(CONNECT_DATA =
(SERVICE_NAME = cwdb)
(INSTANCE_NAME = cwdb)
(FAILOVER_MODE =
(TYPE = SELECT)
(METHOD = BASIC)
(RETRIES = 20)
(DELAY = 15)
)
)
)

CWDB1 =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = TCP)(HOST = xxx.xxx.xxx.x)(PORT = 1521))
)
(CONNECT_DATA =
(SERVICE_NAME = cwdb)
(INSTANCE_NAME = cwdb)
)
)

While I'm selecting data with SQL/Plus on cwap2, DB server changes. The select sentence of connected SQL*Plus can continuously fatch the data. But if I'm updating the data, the session failed.
Is it the correct action of TAF? Does TAF support an update transation's
failover?
If does please teach me the parameter setting and give me same information about TAF.
Thanks.