The program output is intended to be parsed by a script working on the log file or correlated in a spreadsheet. Use the "csv" format (you will have to research "what" is a "csv" format! - use one line in this format per record), including the local timestamp, hostname, remote timestamp and load. The order, and fine details of these entries are up to you - think about what you would want if you were writing the parser to process this file.
The local-timestamp should come from the local machine (the host the script is running on), the remaining values (including remote hostname and remote timestamp) should come from snmp retrieved data.
The Perl program should read the hosts to parse from a simple list in a separate configuration file - the exact format of which is up to you. Make sure you describe the format of this file in your perl script (as comments WITH examples) - no need to supply the file. Remember that host names, community strings (for v2 access), passwords/usernames (for v3 access) and other access details will need to be stored for each server to be accessed - they may not necessarily be the same.snmp is inherently insecure as it potentially exposes a huge amount of sensitive information about a host. As the above will be the only data required, a new snmpd.conf file (for Fedora 14) is required that exposes ONLY the minimum details to provide the requested data. Targets will use readonly access using either of v2c or snmp v3 (username/password) - BOTH must be configured.