Re: Predicting Growth.

Brian Peasland wrote:
> Ben wrote:
> > Ent Ed. 9.2.0.5, DMT's on AIX5L
> >
> > There isn't really an accurate way to estimate growth of a database, is
> > there? I have a history of six months of my data file sizes, but even
> > with that there are too many variables envolved to determine that my
> > database is going to continue to grow at a rate of 50 gig per year.
> >
> > Even if you assume a constant rate of growth, how would you base your
> > projection?
> >
> > For example, lets say my datafiles grew 25G over the last six months.
> > At the beginning of that six months my database was 100G that is 25%
> > growth over the past six months.
> > Now looking ahead for a year. My database is now 125G, 25% of that is
> > 31.25G. Then six months down the road the database will be 156.25G, 25%
> > of that is 39.06G, etc etc. But that original 25% figure was calculated
> > from the 100G incarnation of the database. So that won't work.
> > Do I then assume that my database will grow at a constant size instead
> > of %? So my 125G db will be 175G by this time next year. This seems
> > more accurate, but I still think there are just too many unknowns to
> > actually come to an accurate answer.
> >
>
> The more data points you have, the better to accurately predict your
> trend. I put such data points into Excel and create a chart. I then add
> a trendline to the chart to see where my growth is going to go. You
> might want to consider that.
>
> Also, I'm not sure I quite agree with your example above. You start with
> a 100GB database that grows 25% in 6 months, which gives us 125GB. Why
> do you then assume that that you will have 25% growth, compounded every
> 6 months? Maybe you have linear growth, which means in the next 6
> months, you will grow 25GB for a total of 150GB. By compounding the
> percentange of growth, you are assuming an exponential curve.
>
> That being said, you may have exponential growth...I don't know. It may
> be more linear. Or, it may be a more complicated mathematical function.
> The more data points you have, the easier it is to tell.
>
> HTH,
> Brian
>
>
>
> --
> ===================================================================
>
> Brian Peasland
> dba_at_nospam.peasland.net
> http://www.peasland.net
>
> Remove the "nospam." from the email address to email me.
>
>
> "I can give it to you cheap, quick, and good.
> Now pick two out of the three" - Unknown

I agree. The more data you have, the better. It's kind of like saying "How much capital do I need to retire?" Until you can break out estimates of what you expect your monthly expenses to be, etc., it is really hard.

Depending on how often stats are gathered, you may want to store row counts and row lengths, as that may give you an idea as to how quickly the actual business data volume is growing.

insert into table_size_hist
  select table_name,owner,num_rows,avg_row_len     from dba_tables;

...or something like that to consider.

Regards,

Steve Received on Wed Sep 20 2006 - 14:26:23 CDT