Multiplex Redo logfiles

Hi,
I have a system that has very high I/O load and we run 35 databases on the same server.
All databases is configured with Multiplex redologs and I was considering removing multiplex to help speed up the system and offload the I/O on the system.
I know there are a risk not running multiplex redo logs but is this mainly due to if someone deletes the files or are we talking about possible corruption of the redo logs and loss of data?
As I've never done it before i have to ask:
If I delete the active redolog will the instance stop and become corrupt?
Regards
Morten

Rob_J wrote:
You have good points, EdStevens, but my company are not prepared to take the performance hit. I know that someone could delete them at the OS level, but our SAs know not to touch anything to do with the DBs before asking....If performance is so important than why not to put the redo log files over SSD? The point that Ed has made are very relevant. A hardware RAID won't protect you from human mistakes and this is just a matter of time when you see that a SA is going to come and mention that he didn't know that it was a set of files needed for the database.
If your DBs are log switching rapidly this will add more I/O because of the checkpointing process which has to happen each time there is a log switch. If you have a very large number of small datafiles in your DB, this will increase the time taken to update all the datafile headers, also.The much better way to overcome is this to make VERY large redo logs and control the switching of them using the parameter archive_lag_target parameter. This would solve your issues of keep fiddling the size of the redo logs in varying workloads.
If your DBs are DEV DBs and not PROD ones, and you can restore/refresh them easily, I'd go with one member per log group.
In our DEV DBs we don't run them in archivelog mode either, because we aren't too worried if we lost the DB; we can recreate it easily enough that the positives of having it in archivelog mode don't outweigh the negatives.
If you have optimised everything else, and you are still maxing out the I/O, perhaps you can ask for a hardware upgrade. Also remember that you might be able to tune the statements being run within the DB and look at resource manager to share out resources more evenly.I shall just say that its much easier said than done that you would get a hardware upgrade just like that and also that you would be able to find a db where everything is optimized.
Just my 2 cents.
Aman....

Similar Messages

Multiplexing redo log files

I am using 9i R2 database.
I just want to clear some doubts.
I am having 3 redo log groups with one member each.
Now for mutiplexing redo log files
i want to add one one member to each group so
that each group will have 2 members in some different path.
my current path of log member are
/oracle/files/a1.log (group 1)
/oracle/files/b1.log (group 2)
/oracle/files/c1.log (group 3)
I want to add
/oracle/sysfiles/a2.log (group 1)
/oracle/sysfiles/b2.log (group 2)
/oracle/sysfiles/c2.log (group 3)
A--So is there any need to take database in any mode (nomunt,mount) or i can do it while
database is open ?
I will issue
ALTER DATABASE ADD LOGFILE MEMBER /oracle/sysfiles/a2.log TO GROUP 1 ;
First in inactive group,then
alter system switch logfile; (To make 2 group inactive....and then 3)
B-- All memebers in a redolog group must be of same size? if i am having
100 M.B. of one member(a1.log) is there no need to specify size while adding new member(a2.log)?
Can some one validate and correct my steps or suggest me regarding this?

Hi,
All the solutions are in oracle documentaiton.
See, Members of the same multiplexed redo log group must be the same size (that all member in with in the group ) and Members of different groups can have different sizes. If the same of Files are different resultant would be the with Checkpoints. Since you can not gurantee that Checkpoints will occur at regular internals (Log switch with respect log members). Better to follow the Rules of traffic similarly follow the instrcutions of ORACLE Documentation.. :-)
- Pavan Kumar N

Multiplexing Redo Log and maximum protection mode.

Assume that during writing into redo logs the instance crashes. As a result, members of active redo group are not synchronized, some of them had more data. How Oracle will handle this when instance starts? And there can be case when at startup time some members that had more redo before crash, are lost.
Now assume that we have standby database with maximum protection mode. After LGWR has written to local redo logs and before writing to standby redo logs, primary instance crashes. In this case standby site lost last transaction.
Is it correct? Thanks.

Assume that during writing into redo logs the instance crashes. As a result, members of active redo group are not synchronized, some of them had more data. How Oracle will handle this when instance starts? And there can be case when at startup time some members that had more redo before crash, are lost.
Members of a particular group are written concurrently by LGWR, all members of a log group will have same data. If any member of a particular group is lost or not reachable, oracle will read from the available log member during instance recovery.
Multiplexing Redo Log Files
http://docs.oracle.com/cd/B19306_01/server.102/b14231/onlineredo.htm#i1006249
To answer your second question,In this mode no transaction commits on primary unless the redo is also written on atleast one standby database, otherwise primary will go down.
Check below
Maximum Protection
http://docs.oracle.com/cd/B28359_01/server.111/b28294/protection.htm#CHDHFHJI

Log_file_name_convert , db_recovery_area and multiplexed redo logs

Hi, I have a confusion here, need help.
I have multiplexed redo log members on both asm diskgroup DATA, FRA.
My log_file_name_convert='+DATA', '+DATA'
and db_recovery_area ='+FRA'
db_create_online_log_dest_1=+DAT
Are above parameters right?

log_file_name_convert is simply a string replacing function.Its same behaviour like DB_FILE_NAME_CONVERT
So all entries of <first> will be replaced by <second> and the <third> entry will be replaced by the <fourth> and so on.Yes the strings should be in even numbers like 2,4,6,8.. ther should be sequence for the source & destination diskgroup and so on.
So if you would have:
'DATA','REPLACE','FRA','REPFRA'
all occurences of DATA will be replaced with RAPLACE. But this would also mean ORADATAGRAM would be replaced to ORAREPLACEGRAM.If any ORL's / SRL's are created in DATA in source, by mentioning LOG_FILE_NAME_CONVERT those will be created to the connected proceeding string...
Can you point out please where is mistake in my previous post?

Multiplexing Redo Log Files question

If you are running RAC on ASM on a RAID system, is this required? We are using an HP autoraid which mirrors at the block level and in the documentation about Multiplexing Redo Log Files it says that you do it to protect against media failure. The autoraid that we are using gives us multiple levels of redundancy against media failure so I was wondering if Multiplexing would be adding more overhead than is needed. Thanks for your input.

ASM is quite compex and I'm not going to outline all the advantages or reasons for ASM, but under ASM you can drop and add devices to maintain your capacity needs online without loosing data, which you cannot do using RAID, which requires a re-initialize, for example, regardless of redundancy. Please see the documentation. ASM, like pretty much everything Oracle will add complexity and you will have to check your requirements. ASM is however pretty much the standard. If you use external RAID, make sure your storage is not using RAID 5 or 0. Regarding logical errors, you could for example overwrite or delete a file by mistake, in which case file redundancy does not protect you. If you are looking for reasons or ways not to use ASM, I'm sure you will find them, but what's the point?

Why the redo.logfile in my standby db flash_recovery\dbsid\backupset\ ?

1)I use rman script"duplicate target database for standby nofilenamecheck dorecover"
why the redo logfile automatcly generate in my standby db flash_recovery_area\sid\backupset?while it is in oradata\sid fold on primary role databse?
2)does dataguard apply archivedlog transmitted from primary db directly?or by standby redo log or standby log?
3)How and why standby redo logfile used?

the sequence# 1 has the fisrt_change# greater than
the next_chage#, it's any reason for that?.
SEQUENCE# FIRST_CHANGE# NEXT_CHANGE#
1 534907 551724
I was confused?????
SQL> select 534907-551724 from dual;
534907-551724
-16817

Db_writer_processes and the location of the redo logfile

Is any beneficiary relationship between the number of the DBWR and the location of the redo logfiles? i.e if I spread out the redo logfile to several disks, will that benefit system with multiple DBWR? My system is 10g r2 on AIX 5.3 L with 4 CPU and 4 HDs (73.6 G each).
Message was edited by:
user508054

Is any beneficiary relationship between the number of
the DBWR and the location of the redo logfiles? i.e
if I spread out the redo logfile to several disks,
will that benefit system with multiple DBWR? NO. DBWR writes modified db buffers from the buffer cache to disk (datafiles). So multiple DBWn helps speed up the process (not for Redo Logs)
One of the benefits you get from spreading out the redo logfile to several disks is easier recoverability against damage(with different log members on different disks). This is because LGWR (not DBWR) writes the same redo information to the multiple logfiles you greate.

Why multiplex redo log group ?

Hello,
Why should we multiplex redo log groups if we have only file system which is already mirrored. Is there any one who had an incident when he has only one redo log member per group placed on a mirrored file system but still got corrupted and he felt better to have multiple members even if they reside on the same file system (which is mirrored)
Thanks
Salman

Ansiktet wrote:
EdStevens wrote:
Salman Qureshi wrote:
Hello,
Why should we multiplex redo log groups if we have only file system which is already mirrored. Is there any one who had an incident when he has only one redo log member per group placed on a mirrored file system but still got corrupted and he felt better to have multiple members even if they reside on the same file system (which is mirrored)
Thanks
SalmanThe mirror won't protect you from an SA who deletes '/u01/oradata/redo01.log' because he is running out of space on /u01 and figures its safe to delete a log file.
Or similar types of errors.
The redo and control files are simply too critical to put all your eggs in one basket when planning their protection.:) Thats why you should not use Oracle default .log on redo, instead .dbf or .dbl can be used. THat's why I use the older (pre-10g) default of .rdo for redo logs
However, has this happend anyone for real, that a SA delete Oracle files? Where do you think I came up with the example?
HOw about an SA (or maybe the kind of "fresher" we often see here, Taking a look at a "log" file with notepad?
If he is isnt stupid he should know that Oracle resides on /u01 partion (for example), and should not delete files there without consulting the DBA."should" is the operative word there. There is no accounting for corporate cultures and attitudes.
But end the end, my example was to illustrate that not all problems with redo and control files are mitigated with disk mirroring.

ASM Environment multiplexed redo logs

When we have multi diskgroups like ('+DATA','+FLASH) and configured to multiplexed redologs on asm environment what would be the performance impact while logwriter writing on multiple locations? The diskgroups have different number disks on each diskgroup. Like 100 vs 10 disks on diskgroup.

http://download.oracle.com/docs/cd/B19306_01/server.102/b14231/onlineredo.htm
Multiplexing Redo Log Files help protect against a failure involving the redo log itself.
after you implement, you should check AWR report about waiting.
On ASM, you don't need to make multiple redo files. Just place it (redo file) in good i/o diskgroup (RAID1/0)

REDO LOGFILES Backup

Actually What are Redo logfiles .........?
HOw can i take backup of Redo logfiles from SAP4.7 and ecc6.0 ............?
And in Real time scenario HOw often do we take backup of Redo logfiles.........?

Hi Gibbs,
Redolog files that will be containing the change log for database update.
You can take backup of redolog files using BRTOOLS for ecc6 and SAPDBA for 4.7e.
Normally it should be taken regularly to avoid archive stuck.Better you schedule it regularly using T-code DB13.
check this link
http://help.sap.com/saphelp_nw70/helpdata/EN/8c/09503e30a9d549e10000000a114084/content.htm
Regards
Ashok

Multiplex Redo Logs and Control File

I am wanting to setup an existing Oracle Express 10g instance to multiplex the redo log files and the control file.
Instance is using Oracle-Managed Files and the Flash Recovery Area.
With these options being used what are the steps required to setup multiplexing?
I tried setting the DB_CREATE_ONLINE_LOG_DEST_1 and DB_CREATE_ONLINE_LOG_DEST_2 parameters but this doesn't appear to have worked (I even bounced the db instance).
BTW, the DB_CREATE_FILE_DEST is set to null and the DB_RECOVERY_FILE_DEST is set to the flash recovery area.
Any help is much appreciated.
Regards, Sheila

Thanks for this. My instance originally had two log groups so I've added a new member to each group into the same flash recovery area directory, but have assigned a name. Is this why when I query v$logfile the is_recovery_dest_file is set to NO? Is it ok to assign a name & directory and if not, how do you add a new memeber and allow Oracle-Managed files to name them?
Also, how can I check that the multiplexing is working (ie the database is writing to both sets of files)?
Thanks again.

Multiplexing redo logs and control files to a separate diskgroup

General question this one...
I've been using ASM for a few years now and have always installed a new system with 3 diskgroups
+DATA - for datafiles, control files, redo logs
+FRA - for achive logs, flash recovery. RMAN backup
Those I guess are the standards, but I've always created an extra (very small) diskgroup, called +ONLINE where I keep multiplexed copies of the redo logs and control files.
My reasoning behind this is that if there are any issues with the +DATA diskgroup, the redo logs and control files can still be accessed.
In the olden days (all those 5 years ago!), on local storage, this was important, but is it still important now? With all the striping and mirroring going on (both at ASM and RAID level), am I just being overtly paranoid? Does this additional +ONLINE diskgroup actually hamper performance? (with dual write overheads that are not necessary)
Thoughts?

Some of the decision will probably depend on your specific environment's data activity, volume, and throughput.
Something to remember is that redo logs are sequential write, which benefit from a lower RAID overhead (RAID-10, 2 writes per IOP vs RAID-5, 4 writes per IOP). RAID-10 is often not cost-effective for the data portion of a database. If your database is OLTP with a high volume of random reads/writes, you're potentially hurting redo throughput by creating contention on the disks sharing data and redo. Again, that depends entirely on what you're seeing in terms of wait events. A low volume database would probably not experience any noticeable degraded performance.
In my environment, I have RAID-5 and RAID-10 available, and since the RAID-10 requirement from a capacity perspective for redo is very low, it makes sense to create 2 diskgroups for online redo, separate from DATA, and separate from each other. This way, we don't need to be concerned with DATA transactions impacting REDO performance, and vice versa, and we still maintain redo redundancy.
In my opinion, you can't be too paranoid. :)
Good luck!
K

How to setup redo logfile

I have Oracle 10g on Windows Server 2003. I need to set my redo log file to use 200MB to each group. Also, I need to setup 10groups. Need your assistance.

Dear mrp!
Please don't mind if I ask you about your knowlege about redo logs? Seems that you don't know much about it so I would suggest you to read the Oracle Onlinedocumentation about this topic. Never the less the answer to your question is not as easy as you might think.
1.) Every Redo Log Group consists of one or more memberbs. Oracle recommends to have a minimum of two memberms in each group. So my question to you is: Should every member have a size of 200M or should the whole group have a size of 200M?
2.) You told us that you want to configure 10 Groups. That would generate a very big of head (10 x 200 M of reado logs). Do you really really need 10 groups?
3.) Oracle suggest to multiplex the members of a group to more than one drive for security reasons. Have you thought about this?
Please answer these questions and read ab out online redo logs in the Oracle documentation. After that I may be able to help you.
Yours sincerely
Florian W.

Best way to resise redo logfiles?!!

Hi,
I have Oracle 10g installed on AIX 5.2, and when I was checking average timing of logs switches; I found that a log switch is happening every 179 min.; which, as per Oracle, is too long interval for a log switch to occur (it should be between 20-30 min.). Size of the redo logs is not big, it is just 10MB, but amount of transactions on the database are very small.
I want your advice on what is the best size for the redo logs in my case. In addition, what is the best way to resize them without interrupting database or users?
Regards,

Hi,
but it is not the recommendation by OracleThat's right, but that's right too it's already a small size 10M. If you switch every 180min, that's because the activity is very small, so if you lose your db, the data loss will be very small.
It's the good balance to find, not always easy.
Nicolas.

DB12 - Overview of REDO logfiles - Not yetbackedup - 300 Maximum

Experts...............
In DB12..
Under Redo Log Backups
on Overview of redo log files
....it shows maximum ...300.....
but actually my "oraarch" directory contains 300+ archives , but it still shows "300" as highest number in DB12...
How we can fix and get the right number of redo logs not backed up when archive log count is more than 300..
Rgds

Closing it as no replied and to minimize my open questions
Rgds

Multiplex Redo logfiles

Similar Messages

Maybe you are looking for