Re: log buffer growing and sga tunning

>Perfectly correct. Way too high. This may lead to bad
performance.

Well, many DBA papers recommends 0 or 1 at most redo log space requests - so I raised log_buffer to trying to mantain most log activity on memory - but it failed, and at end I only get a lot of redo wastage. So, I will return to 1 Mb log buffer. Another colleague sugests to use more redo log files - today I'm using 3 of them, with 80 Mb each. I will try 6 of them, wich 90 mb each, and I will add more redo writers (not slaves - I'm in 7.3 here).

> Why don't use archiving. Is this database not important for
you?

I was, but the db was generating MANY hundreds of Mbs in disk - strange, because in v$loghist I was getting only 2 or 3 switchs per hour. Anyway, let's see what happens with the changes.

> How is your redo log configuration? Are they on separate
disks? Is Raid 5 used, are the disks mirrored? You should look at these things?

Well, as I said, I have only 3 disks : a local one and 2 logical disks mounted in RAID-5. So, the only choice was : put the files with heavy reading (i.e., data) in RAID and the files with many writes (log files, control files) in the local disk. I'm fighting for another disk controller, to mount another disk as non-raid, but meanwhile I'm just stucked.

>And besides are you having any performance problems or are you
only concerned about the statistics?

The system is very new - just some months ago. So, we don't have a performance parameter . I'm just trying to put it in the best shape possible (in the near future, the system will get a higher demand).

> To help you any further - you could send me an utlbstat report

Thanks for your good will, Stefan. After the changes I will generate a bstat and send to the list.

[]s

        Chiappa

Got questions? Get answers over the phone at Keen.com. Up to 100 minutes free!
http://www.keen.com Received on Fri Aug 04 2000 - 00:00:00 CDT