ISSUES WITH L.STANDBY DB 10g Enterprise Edition R.10.2.0.4.0 -64bi

Question

Hello,
tnanks in advance for your replies.
I have a logical standby 10.4 Oracle Db in Hp server with 6 cpu's.
It is not 100% a question,but better a discussion.
As far as from the documentation,i try to issue the whole scenario in 3 steps.
1.configure the parameters for log miner,...
apply server,prepare server,sga,eagger,max transaction cout,max servers.Which is the best approach.I have to mention that the problem is if i have a heavy load in primary one and stays 3-4 hours back.Any advice,considerations.
2.Take statistics every day.
3.Make indexes for updates-inserts,cause of the particular way of work tha log miner works.Any suggestion.
Is there something that i have missed??
Thnanks one more time..
http://greekoraclebi.blogspot.com/
*************************

Answer 1

**I am not willing or even trying to like you say "cover" someone in the OTN threads.
Sorry.
**If you want to decrease the gap sequence, simply increase the bandwidth between two sites. When the system is powerful enough to send the archivelogs faster than the Oracle will be fast enough to support huge number of archivelogs generated. Please also bear in mind that there is only one MRP process which applies the transferred archivelogs to the standby system and dealing with those fancy stuff written in the documentation will only help you a little bit but the important part is the line bandwidth.
The Oracle documentation does not calculate the bandwidth needed and it only says "increase it to an higher and higher value".
I have check this from the
SELECT FILE_NAME, SEQUENCE# AS SEQ#, FIRST_CHANGE# AS FCHANGE#,
NEXT_CHANGE# AS NCHANGE#, TIMESTAMP,
DICT_BEGIN AS BEG, DICT_END AS END,
THREAD# AS THR#,applied FROM DBA_LOGSTDBY_LOG
ORDER BY applied,SEQUENCE#;
In heavy loads , i always have 4-5 which in column applied = current and 4-5 which are in 'no',so i have enough logs.
**The next step is the power of the applying process (MRP).To conclude what i am trying to anticipate here, the Oracle documentation parameters are the in lieu of resources on the system. You need to have a lot of CPUs in order to increase that parameter. Also you need to have a lot of CPUs in order to run parallel processing SQLs and other gimmicks.
I can increase it up to 40 servers.Oracle suggest to put 3* cpu's,so in my situation,18.Which 16 are the appliers and 3 the preparers.I have set them and to 40 but the gap was larging.I beleive that it must be a balance.
**From
select min(pct_applied) pct_applied_min
, max(pct_applied) pct_applied_max
, avg(pct_applied) pct_applied_avg
, count(server_id) number_of_appliers
from ( select server_id
, (greatest(nvl(s.total_assigned,0),0.00000001) /
greatest(nvl(c.total_assigned,1),1)) * 100 pct_applied
from v$streams_apply_server s
, v$streams_apply_coordinator c)
**From select name,(least(max_sga,bytes)/max_sga) * 100 pct_utilization
from ( select *
from v$sgastat
where name = 'Logminer LCR c'
, (select value*(1024*1024) max_sga
from dba_logstdby_parameters
where name = 'MAX_SGA'
The percent is 2-3%,so i am ok.
**I have no bytes paged out
**I have no archives gap...
SELECT THREAD#, SEQUENCE#, FILE_NAME FROM DBA_LOGSTDBY_LOG L
WHERE NEXT_CHANGE# NOT IN
(SELECT FIRST_CHANGE# FROM DBA_LOGSTDBY_LOG WHERE L.THREAD# = THREAD#)
ORDER BY THREAD#,SEQUENCE#;
**From this query
select operation, options, object_name, cost,CPU_COST,IO_COST
from v$sql_plan, v$session, v$logstdby
where v$sql_plan.hash_value = v$session.sql_hash_value
and v$session.serial# = v$logstdby.serial#
and v$logstdby.status_code=16113;
i check in real time if i have full table scans.
Here the results are dissapointed.I see many full tables scan(almost 200 in some cases),for updates or delete....
That's why i asked you,if the only solution is the code.
put rely on tables on production,and if possible to put indexes in standby.
I have found some parameter,
EAGERSIZE,what's it purpose .after of test,i concluded that the appropriate value for me is about 1000-1500.
I need PRESERVE_COMMIT_ORDER ,for my data.The data are really sensitive.
Any suggestions?
http://greekoraclebi.blogspot.com/
*************************