i'm taking a stab in the dark here, but if you're using SNMP to fetch data and write a file, then using another script to parse the file, you'd be better off doing the parsing IN the original 'fetch' script and eliminating the intermediate file.

with a large environment, you'll probably need to think about run-time, how often the script is run, and consider using fork() for parallel processing. there is an asyncronous polling feature in SNMP.pm. it might help also.

University of Californa at Davis SNMP includes command-line utilities, MIB queries by name, and a Tk/Perl MIB browser.   From a scripting standpoint, I understand that SNMP.pm combined with SNMP::MIB::Compiler does most of that as well.   Not clear on this, but if SNMP.pm and UCD-SNMP both installed, they might work together somehow.

Actually, I was going to tack parsing onto existing collection script, and hadn't yet realized could skip the intermediate file.   Learning better ways to do it (like this) is a big part of my Monasterial compulsion.   {grin}   I'm always amazed to find that my long-pondered posts turn out to be asking the *wrong* question, yet I still come away with a revelation.
    cheers,
    Don
    striving for Perl Adept

There are a variety of SNMP modules available for perl. I use the same combo you do, and it has full access to the ucd-snmp library calls. Given the motivation, you can easily rewrite all the utilities that cam with the SNMP library in perl.

I have written a script much like what you, ybic. Message me with an email address and I will send you some of the interesting snippets where I gather the same kind of data. Not that my code is anything marvelous, but it sometimes helps seeing what other people have done. Sorry, I am concerned that certain NDAs and copyrights prevent me from slamming the entire thing into the Catacombs.

mikfire