Peformance Data in a Database

Druzhinin Eugene druzhinin at rusmedia.ru
Tue Apr 20 08:43:53 CEST 2004


Hello Ben.

Thank you for your work.
Sorry, it took me qite a long time to realize your suggestions in details.
I will try to modify the patch. But I'm not a good programmer...

May be you look for an instrument like this http://rtg.sourceforge.net/ ?

-------------------------------------
On Mon, 19 Apr 2004 17:52:25 +0100
Ben Clewett <Ben.Clewett at roadrunner.uk.com> wrote:

> I'll get some CGI graphs working when I have a moment.  I must stress, 
> this is really for my own needs to process data into a binary form 
> suitable for fast analysis.  If anybody else finds it useful, or wants 
> to make this a real project, please go ahead.
> 
> 
> At the same time I saw your posting, so let me reply.
> 
> 
> I totally agree with the 'host/service/perfdata_variable'.  I have used 
> it throughout my code, although I have called it 'host/service/metric'.
> 
> I take your point that a given metric has a bad name.  'space' or 
> 'time', or '/usr'.  It cannot be understood without the 
> service_description.  The 'unit' is often not passed.  The 
> service_description is sometimes badly named.
> 
> However, I don't fully understand your next comment.  Do you mean we 
> need a separate table of 'good' or 'approved' variable types onto which 
> we should map the given ones?  May be converting where we can?  Is this 
> what you mean by a MIB?  This sounds like a very good idea.  Although 
> some work will be required.
> 
> I don't entirely agree that my structure has no meaning.  The variable 
> does have a unit.  (s, ms, B, MB, us etc)  It's also linked to the 
> service description.  This does give a human-understanding to the 
> output.  Although I take you point that there might be more that can be 
> done to make this machine understandable.  If there is any need for this?
> 
> Your right, I am dropping all non-numeric data. There isn't a lot which 
> cannot be expressed with a number.  Especially as this data is supposed 
> to be scalar!
> 
> There is a long way to go anyway, and thanks for your comments.
> 
> Ben.
......................

Best regards.
Druzhinin Eugene.


-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click




More information about the Developers mailing list