Chapter 16 : Managing Undo.
Somehow I think this will be a hard chapter. I just had 2 very easy chapters about Oracle monitoring and can’t seem to find the courage to start with Managing Undo. It can’t be that hard can it?
That makes me wonder why I find it a hard subject. ( I shiver when i hear the word undo )
My relationship with undo hasn’t been a good one ( insert bad childhood joke here 
), especially the ora-1555 has been especially cruel to me. At my previous job we often had the problem and each time I thought I had found the solution, it happened again. I added undo space, put the retention longer, … Nothing seemed to work.
I had several questions about the usage of the undo space and never could get clear answers. What is the ora-1555 error? How do I solve it permenantly? By reading the certification book and understanding the inner mechanics of an oracle database I already have a clearer view:
Ora-1555 is the error that is used when the database looses the ability to give other queries a consistent view of the database. The undo space keeps the “old records” so that query’s running while the data is (being) changed and is uncommitted still see the old data. When a database can’t guarantee this anymore, the running query is cancelled with ora-1555. the retention period and size all determine how long that is. Size is the most important one I guess. The new undo advisor in 10g should be able to predict the undo space needed because that was a real mess.
Now let’s study chapter 16 and see if that is correct
