Re: Restoring & recovering in NOARCHIVELOG

Quarkman <quarkman_at_myrealbox.com> wrote in message news:<oprsu967tir9lm4d_at_haydn>...
> On 25 Jul 2003 05:39:39 -0700, Pete's <empete2000_at_yahoo.com> wrote:
>
> > and you'll have to have all of the archives that
> > were created throughout the day. The size of the redo log does not
> > necessarily matter, however, you don't want to be switching logs every
> > minute, once every 15min is good.
>
> No it isn't. Log switch when its suitable to do so. Saying that every 15
> minutes is 'good' is a bit like saying 'I weigh 140 pounds... do I look
> good in blue?'. It's meaningless, in other words. It's good for you, but
> not for everyone else.

The problem is in defining suitable. The best thing, of course, is to have management specify a service level agreement, but at many sites that implies a huge disaster-recovery planning project. And if you simply ask what level of granularity, they automatically say they need up-to-the-second recovery - even if the thing is only being updated nightly with informational data and transactions are not really that critical. So you tell them how many kilodollars that will cost, and use a rule-of-thumb like 15 or 30 minutes when they blanch and tell you they'll get back to you. That level of granularity is reasonable for a small/medium non-24 system given how long it takes to respond to an outage without doubling hardware.

jg

--
@home.com is bogus.
http://www.signonsandiego.com/news/uniontrib/fri/news/news_1n25camacho.html