I have 3 network mgmt Q’s:
1) I am looking for inexpensive network management s/w to tell me when my core network gear is down: APs, cisco router, cisco switch, and dsl modem.
2) I need a s/w tool to be able to track monthly GBs of network throughput by customer. It needs to track by MAC address since I do dynamic IP addressing.
3) I would also like to be able to remotely manage and reboot the APs, cisco router, cisco switch, and dsl modem when necessary.
ANY ideas on any of these questions would be greatly appreciated.
I have BAM installed for bandwidth shaping and it works great.
Prism?
I’m using a combination of Cacti and Nagios. I haven’t set them up to my full satisfaction yet, but it looks very promising in terms of featureset and cost.
I’m not familiar with Cacti and Nagios (or Prism) … I’ll look them up on the web. What do each of them do? Do you know approx. cost?
thx
- Nagios. Its GPL, enough said
2) MRTG. Its GPL, enough said
3) For that I recomend sticking to whatever the equipment provides for remote management. IE: CLI via telnet for Cisco, web interface for Canopy, whatever your DSLAMs use (CLI via telnet on my Paradyne stuff).
Some form of SNMP utility would probally work for some configuration changes. Call me paranoid, but enabling write access on SNMP is a little concerning to me.
The problem with MRTG is that it is fairly heavy on system resource. FYI.
Shaman wrote: The problem with MRTG is that it is fairly heavy on system resource. FYI.
not if you use RRDTool integration
JFFNMF offers everything you need:
- Alarm Monitoring (via SNMP traps)
- Traffic / Bandwidth traffic, utilization, losses drops, bla, bla, … (with nice charts)
- Grouping of elements per zones, customers.
We’re using it for about 2 months. Works like charm.
Maybe a dumb question, what is JFFNMF?
Sorry.
Aaron
JFFNMS. Sorry!
Just For Fun Network Management System 
Cool, thanks.
Aaron
Any special srcipts for JFFNMS?
FYI: in the Cacti forums, someone has posted Canopy scripts + definition files which allow you to collect data on jitter, rssi, # of SMs, etc. I haven’t had a chance to try them out yet, but they look promising.
Shaman wrote: The problem with MRTG is that it is fairly heavy on system resource. FYI.
That was the case in 1996. Today things have changed. I am graphing 30 SMs on the head node of my squid cluster using MRTG and the basic example scripts. Jitter, RSSI, ethernet interface, and rf interface.
The same unit is also graphing all interfaces on all routers in our internal network (4 routers), all external interfaces on our gateway routers (2 routers), and all of the circuits upon which we backhaul our aDSL subscribers (2 routers, 4 circuits).
I should mention that it is also switching all the traffic in our squid cluster between users and the internet. I think the physical cache nodes are maybe 60 or 80 gigs each and there are two of them, I'm not sure as it's been a while since I built the actual proxies themselves.
Unit is a Linux based, HP Vectra with a 333 MHz processor and 256 M ram, and some tweaks. The P-III, 700 MHz box that had been doing the job died a month or so ago and I just swapped the raided drives into the old Vectra and turned it on.
MRTG might have been a load at one time, but it isn't showing up on that old HP with enough load to matter today.
[/quote:1svgzurt]
micers wrote: [quote="Shaman":1svgzurt]MRTG might have been a load at one time, but it isn't showing up on that old HP with enough load to matter today.
I was generating over 2,500 graphs + three levels of dates (that's more than 7,500 total for the math-challenged) every five minutes. The core forwarding switch alone generates 336 graphs (4x1GoF,96100TX,env,vlans). That's *without* adding the Canopy network.
Yes I could have had MRTG store the data with rrdtool, but then I'd have to get some complicated and hard to use program to browse through them all, instead of my nice custom-made, notated web pages. MRTG doesn't generate any graphs (on its own) when using rrdtool storage.
Cacti is much easier, and it has two things mrtg should have had years ago: 95th percentile and total bandwidth calculations on the graphs. I can scan a whole switch and have it added to a view tree in minutes. The only thing I haven't yet come to terms with is that my Canopy stuff is on two separate VLANs that the web server running Cacti can't reach... so I may have to tunnel to those networks if I want to have my network view all in one application.
Just for fun, I timed the Cacti poller script on my machine (dual 2.4Ghz P4 w/3.5G mem, 15K SCSI drives, runs qmail on Linux as well) and came up with:
real 0m11.609s
user 0m0.894s
sys 0m0.800s
http://www.cacti.net/
http://forums.cacti.net/viewtopic.php?t=4345&highlight=
I have an alterior motive in recommending Cacti... the more eyes on it, the better the package will become, IMHO.