[Gllug] Predicting using mrtg data

[Tethys]

Paul Nasrat writes:

>Can I make the assumption that if I expect a to be serving 5 times the
>amount of stuff that I am serving now on a linearly distributed basis I
>can take each snapshot measurment and factor it up by 5?

Sounds reasonable.

>I'm quite scared by the peak numbers atm :)

You shouldn't necessarily be. If the peak is 10x the average, then you've
just got some irregular usage patterns, and you don't need to accomodate
your peak traffic -- just accept that you'll max out your bandwidth at
peak times. If, on the other hand, your peak is only 1.5x average, then
you need to allow for it in your bandwidth calculations.

Tet

-- 
Gllug mailing list  -  Gllug at linux.co.uk
http://list.ftech.net/mailman/listinfo/gllug