Problems with Graph Output for Gigabit Interfaces

trkemsit · Post by **trkemsit** » Fri Oct 29, 2004 8:06 am

Hey guys,
Second post i have abou tthis topic. Now i have searched this forum extensively. I am actually trying to finish my setup on a Win2k Server with Apache versus what i have no which is a Debian Linux Solution. (I want to look at WMI graphs).

Currently i have been searching for the last month to figure out why my Gigabit interfaces on my campus network looks so strange

I have used 64Bit counters and have tried adjusting the polling intervals from 5 minutes to 2 minutes. This is what i normally get. Below. By the way here are my data sources:
/usr/bin/rrdtool create \
/var/www/html/cacti/rra/mia_6k_mdf1_traffic_in_9.rrd \
--step 300 \
DS:traffic_in:COUNTER:600:0:125000000 \
DS:traffic_out:COUNTER:600:0:125000000 \
RRA:AVERAGE:0.5:1:600 \
RRA:AVERAGE:0.5:6:700 \
RRA:AVERAGE:0.5:24:775 \
RRA:AVERAGE:0.5:288:797 \
RRA:MAX:0.5:1:600 \
RRA:MAX:0.5:6:700 \
RRA:MAX:0.5:24:775 \
RRA:MAX:0.5:288:797 \
RRA:LAST:0.5:1:600 \
RRA:LAST:0.5:6:700 \
RRA:LAST:0.5:24:775 \
RRA:LAST:0.5:288:797 \

After readiny many people post about maximum Values i have adjusted those between 125000000 and 1000000 and also 1000000000 and also 1000000 and also 100000. Same results (each time deleting the rra files!)

Post by **raX** » Sun Oct 31, 2004 10:37 pm

Forgive me if I missed something in your post, but are you referring to the two outbound spikes in your graph as strange? Have you confirmed that these two spikes aren't representative of actual traffic on your network?

Either way, 64-bit vs. 32-bit counters shouldn't make a difference here as 32-bit is good up until around 114 mbit/sec. The most common cause of traffic spikes (aside from actual traffic) are from counter resets caused by a device reboot. Did this device reboot at all during this graph's timeframe?

-Ian

trkemsit · Post by **trkemsit** » Mon Nov 01, 2004 7:54 am

Actually the problem is that i know that we should be doing well into the 100MB of traffic, atleast at some points of the day because we have multi 100MB files transfering at all times for PACS images, should all of my gigabit links to the distribution/core be under 100mb? Its very strange.

Post by **raX** » Mon Nov 01, 2004 2:26 pm

Just because you are transferring multiple hundred megabyte files during the day does not necessarily mean that your average will be that high. They daily graphs represent a five minute average, so even if you transfered a 300MB file it would only register as 1MB (8mbit) on the graph.

If using 64-bit counters produced the same results, I would be reluctant to challenge the numbers. If you are still in doubt however, I suggest using a separate tool (like MRTG) to compare with.

-Ian

trkemsit · Post by **trkemsit** » Mon Nov 01, 2004 7:41 pm

Hmm interesting so just because we have a lot traffic possibly on the network those numbers may actually be accurate... Frightening

Post by **raX** » Mon Nov 01, 2004 8:29 pm

Quite possible. Yet another reason why management doesn't want to use graphs to justify the company's new 10 Gigabit Ethernet investment

.

-Ian

Post by **TheWitness** » Mon Nov 01, 2004 8:32 pm

Aren't you supposted to be studying?

Larry

Post by **raX** » Mon Nov 01, 2004 8:34 pm

TheWitness wrote:Aren't you supposted to be studying?

That was yesterday. Now I'm supposed to be writing unit tests for my traditional life cycle course

.

-Ian

trkemsit · Post by **trkemsit** » Tue Nov 02, 2004 9:46 am

I am just the security guy and backup senior network engineer, but i am thinking that as soon as the 10 modalities for the PACS system come online and also the wireless with VOIP that these numbers should go up

Cacti

Problems with Graph Output for Gigabit Interfaces

Problems with Graph Output for Gigabit Interfaces

Who is online