Help monitor Gigabit Cisco Interfaces

[theRENEGADEkemist]

I know that this question has been posted 10000 times. I've search all over the forums and found many many articles.

Here is the obvious issue:

I do have this graph setup to do In/Out Bits (64-Bit Counter)
This is a Gigabit Fiber Uplink to the core, the other fiber links to another set of switches.

Gigabit Interfaces on my switch are showing probably lower bandwidth than possible. Here is a graph of an example.



Now i have cacti 0.8.6g with all the patches
Windows 2k3. Running CMD.PHP.

Only 8 devices on this.
IIS 6+
PHP 4.0 (or is the latest 5?)
ActiveState Active Perl
The Poller runs every 5 minutes
Mysql 4


I am looking for suggestions. Are 64 Bit counters wrong? Should i slow or speedup the poller? What will that do to WMI based Pollling?

[BSOD2600]

How do you know they are wrong? What do you have to compare against? Sure you're monitoring the correct interface?

[theRENEGADEkemist]

Well... Those interfaces are the uplink interfaces to a set of 3 stackables. One of the interfaces runs into my core and is the Spanning-Tree Root for the other 2 stackables. Over 50 devices are connected to it, they are all actively transfering. There are clinical servers that run on it.


Lastly they are 1000MB connections. Do you not think that only 2.6 MB of maximum traffic is extremely small for those statistics? I can trasfer a 100MB file and not see the difference.

Although this is what i see as counters on that switch:
int gig0/1
5 minute input rate 542000 bits/sec, 233 packets/sec
5 minute output rate 191000 bits/sec, 160 packets/sec

int gig0/2
5 minute input rate 73000 bits/sec, 41 packets/sec
5 minute output rate 264000 bits/sec, 109 packets/sec

Does that makesense to anyone?

Although mathematically 542000Bits = 43MBytes don't iit?

[guerra6880]

Well... Those interfaces are the uplink interfaces to a set of 3 stackables. One of the interfaces runs into my core and is the Spanning-Tree Root for the other 2 stackables. Over 50 devices are connected to it, they are all actively transfering. There are clinical servers that run on it.

Lastly they are 1000MB connections. Do you not think that only 2.6 MB of maximum traffic is extremely small for those statistics? I can trasfer a 100MB file and not see the difference.

Try transfering a much large file so that you can sustain a 5 min average increase, eg. 2 GByte. I have seen 150 people behind a 100 Mbit/sec connectiong only generate 1 or 2 Mbit/sec of traffic on a 5 minute interval.

Although this is what i see as counters on that switch:
int gig0/1
5 minute input rate 542000 bits/sec, 233 packets/sec
5 minute output rate 191000 bits/sec, 160 packets/sec

int gig0/2
5 minute input rate 73000 bits/sec, 41 packets/sec
5 minute output rate 264000 bits/sec, 109 packets/sec

These #s look inline with the graph that you have attached above.
int gi0/1:
542000 bits/sec = 542.0Kbits/sec input
191000 bits/sec = 191.0Kbits/sec output

int gi0/2:
264000 bits/sec = 264.0Kbits/sec input
73000 bit/sec = 73.0Kbits/sec output

Although mathematically 542000Bits = 43MBytes don't iit?

The default interface template records in bits/sec. See the legend on the left side of the image that you post.