Wednesday, February 11, 2009

AWK Programming Language [revisit]

I'm now revisiting AWK language for trace file processing. Awk has been widely used for trace file processing, since (I believe) it is easy to do column filtering.

The column can be represented using $1 for column 1, $2 for column 2 etc.. , using other language like Tcl requires list manipulation which is not as straight as using AWK.

However, of course, the syntax of AWK is a little bit different than Tcl and C.

I will be "reusing" AWK to calculate throughput , loss , jitter etc. There are quite a few AWK codes available in the net for this purpose, however I want to make the code as flexible as possible and cater for our group needs.

First, I will be looking back at RFC's and papers on the network performance formula.Then, get available AWKs from the net (to enhance understanding) and modify the code to suit our need.

Our new research assistant maybe can help me working on this??

Performance of AWK?? Hrmm... anybody have experience using AWKs? How is the speed compared to C and Tcl (or MATLAB?)

I will not focus on performance for now , I believe it is acceptable, for now the main factor is whether it make my work easier or not.

"Live life in this modern world is complex enough! Programming Work Shouldn't Be "


Mohd.Taufiq said...

I think when using awk file we also need to master regular expression, because to search or match patter is more accurate and better then using looping...

zul's said...

nice tip taufiq, tahnk you for sharing it here. Regexp is very powerful indeed if we know how to use it.

Mohd.Taufiq said...

It is very easy, we can know by trying every method to match our searching patten