Tracking server bandwidth usage and other stats

[Lst Recv]

Hi all.  I'd like to set up automatic server bandwidth/network, CPU,
and memory tracking.  I know there are a lot of products
(mrtg/rrdtool, negios, cricket, zambix?, etc.), and my requirements
are a bit unique, so I'd figure I'd ask you all for some advice.

REQUIREMENTS:
1) Be able to read the data in a text file via SSH (ie, no pretty
pictures, should human readable output)
2) Work without SNMP (it only has to monitor localhost)
3) Track: total bandwidth usage (eg 3.2GB this month)
4) Track: any rates which would indicate a bottleneck, on the net, cpu, or drive

WOULD LIKE:
1) For #1 above to be easily piplinable via unix power tools (ie,
grep, cut, etc.)
2) To be able to send alerts when bottlenecks or limits nearing
3) To be able to track key applications specifically - how much load
they use, how much latency (key apps include apache httpd, mysql, and
qmail)

Any help appreciated.