Re: Freelist Groups in 11g documentation

  • From: William Muriithi <william.muriithi@xxxxxxxxxxxxxxxxxxx>
  • To: "'development@xxxxxxxxxxxxxxxxx'" <development@xxxxxxxxxxxxxxxxx>, "'oracle-l@xxxxxxxxxxxxx'" <oracle-l@xxxxxxxxxxxxx>
  • Date: Sun, 10 Jan 2010 12:20:01 -0600


I believe 10 and 11g uses ASSM by default for segment space management. That 
may be the reason there is little documentation for the two version above.

Look back on discussion between mid December and end of the year and there was 
a lot of discussion on this topic. On a phone at the moment so can not help dig 
but a google for "ASSM oracle-l" should do it.

From that discussion, I think it is not worth managing segment manually. ASSM 
seem to do the job fine. It was recommended a better solution was partitioning 
for data and reorganization - increasing to decreasing - for index. Any way, 
check it up as I am writing this from top of my head and weekend drinking may 
not be helping

----- Original Message -----
From: oracle-l-bounce@xxxxxxxxxxxxx <oracle-l-bounce@xxxxxxxxxxxxx>
To: ORACLE-L <oracle-l@xxxxxxxxxxxxx>
Sent: Sun Jan 10 11:24:19 2010
Subject: Freelist Groups in 11g documentation

Dear list members,

seems like I am beaten by Sunday evening blindness but I can't find
anything useful regarding freelist groups in the 11g documentation. I
queried and google but all I got back were hits in 9i
and OPS documentation.

So it seems users are discouraged from using manual segment space

I am asking because I couldn't reply properly (to my own standards) to a
developer who wanted to increase ITL entries and freelists on some index
segments where he saw lots of buffer busy waits (read by other session)
during a parallel alter table move ... updating global indexes. Oh, the
developer claimed they were in data blocks, not segment headers (based
on an AWR report).

Is there a good source of information available besides said slightly
older documentation? Or does anyone know why you'd increase freelists
and ITL entries for indexes? From a thread on this list I thought there
wasn't much point in doing so...

Any pointers to information appreciated.


PS: Oracle 64bit on RHEL 5.3/3 node cluster (and no
interconnect problem :)

Other related posts: