Hey all,
I am working on a project on Green Computing.What i need to do is store the no. of requests hitting server during in a day in database.That is,in every 15 mins i have to store total hits during that duration in database,so that i can obtain network traffic pattern !!!

I am not able to find a way to do so....

Plz suggest sumthing !!!

Regards
Sanket !!!

Hi -

Assuming you're running Apache, have you considered parsing the http requests log?

Something like (shell example):Also, I'd strongly recommend writing this in something like Perl or Python:

a) scripts are faster/easier for jobs like this
b) Python and Perl have all the flexibility of shell scripts, but they also make it easy to talk to mySQL
c) examples/tutorials for Python and Perl are abundant - you can literally have something up/running in an afternoon (if you've never used them before), or an hour or less (if you already know one or both of these languages).

'Hope that helps .. PSM

I'll suggest you spell out your words.

And if you need this, consider looking into Nagios, or other system-monitoring tools for Linux that are out there. They already providing trending, graphs, etc., and are ready to go.

Also, you don't say what KIND of requests, hitting what kind of server, running what version/distro of Linux, etc.

Hey,

Thanks for your replies....

I am working on Ubuntu....have to do this project on heterogeneous servers...so basically whenever there is some low load on servers,we have to switch some servers to hibernation and switch them back to normal working mode when no. of requests is high...hav to take this decision using historical data .. thats why want the no. of requests hitting any server(webserver,ftp,mail etc etc )....and from requests i mean to say client requests !!!!

I have looked into various tools but none of them provides no of requests hitting any server..they talk about bytes of data transferred,network usage etc...

Plz suggest sumthing !!!

Again SPELL OUT YOUR WORDS. And "client requests" is very broad.

Nagios and other tools measure throughput, but if you want to measure the different protocols, you'll have to install something like MRTG, and write something to monitor all the different network protocols, and graph them.

Or you can follow my original suggestion and "grep" the darned http access log

... and for all services, use the LOG option in iptables and use Perl to pull out nums for each service and put in DB. Note that this is going to use a fair bit of disk space, so be prepared to tweak the logrotate settings