Re: backup and recovery in data warehouse

Yes, I see your point. The more I read the more I realise I should run in archive log mode. And by using read-only tablespaces my weekly cold backup timeframe will be reduced. I'm also suggesting that we run multiple tape drives in parallel, use a robotic arm and a tape library. this will speed up the backups.

I've read a couple of excellant papers 'System Architecture For Oracle 8 VLDBs - A Case Study' by Gaja K. Vaidyanatha of Anderson Consulting (http://www.ioug.org/repository/index.html - iouga99 papers, you need to be a member to access this) and 'Configuration and Management of An Oracle Data Warehouse' by David C.Kreines, Rhodia, Inc (I think this came from an openworld site?)and also 'Riddles of Read_only Tablespaces' by Venkat Chandrasekar. (also from openworld i think).

I'm still trying to figure out what is the best way to define tablespaces for future partitions.

Thanks
Jeremy

In article <389EF4B3.66CBC231_at_wolfenet.com>,   Jeremiah Wilton <jeremiah_at_wolfenet.com> wrote:
> Jeremy,
>
> You say that you will not be able to back up all read/write
tablespaces
> in a single cold backup, but also that you don't want to use
archivelog
> mode. Given these two facts, how would you restore the database in
the
> event it became necessary? Even if you restored separate pieces of
the
> database from separate days, you would have no way to bring them
> consistent with each other. The restored database would be unopenable.
>
> --
> Jeremiah
>
> jeremycdba_at_my-deja.com wrote:
> >
> > Thanks for the replies. To answer a few of the questions raised.
> >
> > This data warehouse will be at least 1.5 TB in size that's why we
can't
> > do nightly cold backups. Having some read-only tablespaces will
reduce
> > the amount to backup. But I expect that we'll never be able to
backup
> > the entire database in one go even excluding the read only
tablespaces.
> > Exports and imports will just be too slow.
> >
> > We're thinking of NOT running in archive log mode. Being a data
> > warehouse users will only be performing reads. The only
updates/inserts
> > will be performed during the loading of data, which can be reloaded
if
> > there are any problems. Having archiving turned off will improve the
> > performance of this. So I can't see the point of running in archive
log
> > mode. Nice in theory I know so this may change in the future.
> >
> > On the tablespace sizing and creation. We're going to be
partitioning
> > by month and each of these partitions will be in a seperate
tablespace.
> > So how do we know how big to make each tablespace and when should
they
> > be created? If we guess at the size and they're too small then we'll
> > have problems, too big and we're wasting space. That's why I
suggested
> > using autoextend. And should I create them all for say two years
before
> > they're needed???
>

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Wed Feb 09 2000 - 16:23:57 CST