Re: 24x7 database

In article <6en6g0$pkb$1_at_nnrp1.dejanews.com>, <niblue_at_yahoo.com> wrote:
>I'd re-iterate Steve's comments - as an Oracle consultant I've been to far too
>many sites where people have tried to put in place a "clever" backup scenario
>(hot backups, no downtime etc) and have got themselves into a situation where
>their database is either unrecoverable, or will take ages to recover.
>
>SAP is a complex bag of shit as well (in database terms at least), so you will
>need to do a lot of work and testing before going down this route. When
>putting in place backup scenaries don't addressing recovery (timescales etc),
>and do recovery testing frequently
>
>People often make the mistake of going 24x7 without adequate staff and
>facilities - make sure it doesn't happen to you, as I've seen people lose
>their jobs when these things go wrong.

Boy, you can say that again! I'll get a horror stories page up one of these days... If I can ever find time...

I'm not convinced Oracle can handle all situations 24x7. For example, I just had a major downtime because someone loaded a bunch of data with SQLload direct, which didn't finish and left an important index that takes many hours to rebuild corrupt. So now I have to come up with a hacked together routine with double tablespaces so they can load one while the other is running, then switch and load again, etc. It probably wouldn't ever happen without direct, but since it _has_ happened, it has contributed to the strange superstitious behavior many shops develop. And I'm not convinced making a more complicated database because of one problem that could be avoided by "don't do that" is a good answer.

It also makes me wonder how people tune databases with mixed requirements - a general case of small transactions coupled with periodic large loads. So far I've just been leaving large log files, but as I implement hot backups I wonder how tricky I should get. It would only take one incident of "oops you did the big load with small log files" or "oops we can't restore that because the log files were too big and hadn't switched yet" to cause yet more superstitious behavior. Of course, the big load stuff is our "static" data - I'm working on splitting the _volatile_ stuff into it's own space.

It all boils down to trusting Oracle to do its job - and the more complex its job, the more subtle the bugs can be.

>
>---Nile
>
>In article <890158371.11764.0.nnrp-02.c2de712e_at_news.demon.co.uk>,
> "Steve Phelan" <stevep_at_XXnospamXX.toneline.demon.co.uk> wrote:
>>
>> One piece of advice: Get some help.
>>
>> If you are looking at a system of this size and complexity (we run a few
>> similar systems) and are asking such basic questions - and no offence, BTW,
>> you have every right to ask - I really think you are out of your depth. The
>> issues surrounding this system will be complex, but nothing that can't be
>> managed. Don't look at a few postings to a newsgroup for your answers.
>>
>> Go away and plan. Get some training and consultancy. Then test. And test
>> again.
>>
>> Sorry to be so hard. But I've seen people fail before in this type of thing
>> and the results aren't pretty.
>>
>> Try reading the Oracle Database Administrators Guide for an overview of
>> hot-backups if you want an intro., but treat it as just that, an intro, not
>> a solution. You might also want to look at Oracle Backup Manager and other
>> thrid party backup tools that integrate with Oracle.
>>
>> One other point, try and separate your 'static' data into read-only
>> tablespaces, then you won't have to back up so much...
>>
>> Later,
>>
>> Steve Phelan.
>>
>> tammy.adler_at_ipaper.com wrote in message <6ek8u5$3gh$1_at_nnrp1.dejanews.com>...
>> >I am interested in hearing from dba's in a 24x7 environment, large
 database.
>> >Currently we have a 12 hour maintenance window each weekend. They are
 going
>> >to shorten this to a 4 hour window 3 weekends of the month, with no outage
 on
>> >the month-end. By fall, it is requested to be true 24x7, with scheduled
>> >outages every 6 months.
>> >
>> >This is a SAP application...the database is 150GB and growing
 rapidly...will
>> >probably double by year's end. The box is the IBM Raven, with AIX 4.3.
>> >
>> >How much housekeeping should I do in advance? What else should we do in
>> >preparation. Can a system be truly 24x7 without any scheduled down-time?
>> >What problems can we expect?
>> >
>> >E-mail or post your replies.
>> >
>> >-----== Posted via Deja News, The Leader in Internet Discussion ==-----
>> >http://www.dejanews.com/ Now offering spam-free web-based newsreading
>> >
>>
>>
>
>
>-----== Posted via Deja News, The Leader in Internet Discussion ==-----
>http://www.dejanews.com/ Now offering spam-free web-based newsreading

-- 
These opinions are my own and not necessarily those of Information Quest
jgarry@eiq.com                           http://www.informationquest.com
http://ourworld.compuserve.com/homepages/joel_garry
"See your DBA?"  I AM the @#%*& DBA!