|
Hi,
[del]
> Load balancing is a feature of piranha (there are multiple algorythms to
choose
> from), but it does not work the way you are suggesting. It does not
> "time" how long a fetch takes. There are options to base the balancing on
> number of connections, a manually specified "weight" (to favor larger
systems
> over smaller ones), and cpu load.
Yep, I am familiar with this (hey, I actually read the how-to before asking
a question and I even set up a test-cluster ;)
> > Has anything like this been implemented yet / are there good reasons
against
> > doing this ?
>
> There are several reasons for not doing it that way, as it is very
subjective
> to the environment rather than server performance. Different web pages are
I think the "dumb" way of measuring would be an advantage here: I do not
care how much CPU/Disk-Load a (webserving) realserver has, as long as it
serves webpages pretty fast (at last thats all the person viewing my
websites cares about). You are right that one should probably measure the
response-time of a costy dynamic page (if you do have dynamic pages on the
cluster) and that the selection of the right page is a important thing for
this to work.
I´m sorry that I forgot about VS-Tun: Of course the suggested type of
measurement would only make sens eif all realservers are on a local network.
> This would also interrupt the real web service activities taking
> place (the act of measurement affecting the thing you are measuring).
I don´t think that this is critical. Requesting a dynamic page every 5secs
or so should not bog down the server so much that you actually "influence
the results by measuring". Besides: Don´t you do that when you measure
CPU-Load also ?
Greetings,
Jochen
|
