cancel
Showing results for 
Search instead for 
Did you mean: 

Log Area filling fast?

Former Member
0 Kudos

I have some questions regarding the log area.

We are running a production database (non-sap-system) that is about 60 GB big (perm size). Log volume is 1 GB.

Database growth is about 100 to 150 MB per day.

But: Autolog crates more than 30 GB of log backups per day (about 1/2 of the perm size and 30 times the growth of the database).

The ration between database growth and the amount of log backups created seems somewhat off the wall. Correct me if i am mistaken and this ratio might be normal.

What factors do affect the filling speed of the log area and amount of log backups respectively.

SQL committed by our app?

Size of data volumes and log volumes?

Usage percentage of data volumes?

Internal optimization of b-trees by maxdb?

Block-size of file system?

And where is the best point to look for reasons for that rapid growth?

Maxdb is version 7.6.05.09 running on Windows Server 2003 x64.

Thanks

Jörn Menkhoff

Accepted Solutions (0)

Answers (1)

Answers (1)

lbreddemann
Active Contributor
0 Kudos

>

> I have some questions regarding the log area.

> We are running a production database (non-sap-system) that is about 60 GB big (perm size). Log volume is 1 GB.

> Database growth is about 100 to 150 MB per day.

> But: Autolog crates more than 30 GB of log backups per day (about 1/2 of the perm size and 30 times the growth of the database).

> The ration between database growth and the amount of log backups created seems somewhat off the wall. Correct me if i am mistaken and this ratio might be normal.

Well, data area usage is rather static - while created log is totally dynamic depending on the data change activities.

> What factors do affect the filling speed of the log area and amount of log backups respectively.

> SQL committed by our app?

There's an important point here. It's not only the committed transactions that fill up log!

If there are many rollbacks this won't change your data area but it does create log.

> Size of data volumes and log volumes?

No, the size of the volumes has no influence on the amount of log data that is created.

> Usage percentage of data volumes?

Nope.

> Internal optimization of b-trees by maxdb?

Again: no.

The log information do not include B*tree manipulation data but only logical data change information (e.g. for this row change column X to value Y).

> Block-size of file system?

Nope.

Besides the fact that you cannot change it the block size has nothing to do with the amount of log written.

> And where is the best point to look for reasons for that rapid growth?

I'd try to get a better picture of the activities of your database.

Make sure to have DBAnalyzer running and check the detail logs (yes, they are difficult to read and interpret ... )

As an simpler approach you may use


info sysddmonitor

in dbmcli to get an overview of the instances workload.

regards,

Lars