Re: Locally Managed Tablespaces ... again!!!

joel-garry_at_home.com (Joel Garry) wrote in message news:<91884734.0301201536.4d9b1d1_at_posting.google.com>...

> > Er, and to what are you referring then. As I say, show me *how* contention
> > can arise between a table and its index. The reads are serialized. The
> > writes are random, in the sense that DBWR is just as likely to flush out two
> > table buffers as a table buffer and its related index buffer. There's no
> > contention anywhere I can see. What's in the 'everything' pot of goodies
> > you're thinking about?
>
> A (write to the table followed by a write to the index)* 100 people at
> the same time.

In this scenario, why would this:

create tablespace t1 datafile '/disk1/t1.dbf'; create tablespace t2 datafile '/disk2/t2.dbf';

create table t ( .... ) storage ( initial 512k next 512k pctincrease 0 ) tablespace t1;

create index i on t( ... ) storage ( initial 256k next 256k pctincrease 0 ) tablespace t2;

be better than this:

create tablespace t1 datafile '/disk1/t1.dbf', '/disk2/t1.dbf';

create table t ( .... ) storage ( initial 512k next 512k pctincrease 0 ) tablespace t1;

create index i on t( ... ) storage ( initial 256k next 256k pctincrease 0 ) tablespace t1;

(or RAID striping)
?

Cheers
Stu Received on Tue Jan 21 2003 - 11:01:07 CST