Re: index contention in RAC

  • From: Hemant K Chitale <hkchital@xxxxxxxxxxxxxx>
  • To: johan.eriksson@xxxxxxxxxxxx, oracle-l@xxxxxxxxxxxxx
  • Date: Fri, 21 Apr 2006 23:39:04 +0800

Try increasing the CACHE size for your Sequence.

At 10:58 PM Friday, you wrote:

create index idx_pk_n1 on t(n1) tablespace index_test logging reverse local /

alter table t add constraint pk_n1 primary key(n1)

Our machines are AMD64 running RHEL 4 and we have 2 nodes in the RAC, storage is ASM, the blocksize of the db is 8K, databaseversion is The usage of the tables will be mostly inserts and not that much querying. At the moment I am testing with approx 200 concurrent users, all just doing inserts (which is the scenario we expect), the clients connects with jdbc (trough hibernate) The tablespaces are locally managed and ASSM.

The problem we have are large amounts of GC buffer busy, since the
primary key is generated by an sequence I have made the index reverse to
eliminate some of the buffer busy events and that helped alot but the
major waiting is still on gc buffer busy and I want to know if there is
more I can do to minimize/eliminate this?

Hemant K Chitale


Other related posts: