Output without Ethernet headers?

Ricardo S super.ismiti at gmail.com
Wed Oct 26 09:49:20 EDT 2011


OK. Thanks Carter.

[]'s
Ricardo.

On Wed, Oct 26, 2011 at 3:33 PM, Carter Bullard <carter at qosient.com> wrote:
> You have to configure argus to provide the appbytes, using the argus.conf file.
> Carter
>
> On Oct 26, 2011, at 9:05 AM, Ricardo S wrote:
>
>> Hi Carter,
>>
>> Thanks for replying. I want to have the total bytes transmitted in
>> unidirectional flows to calculate the total bandwidth used in a period
>> of time. So, I tested the appbytes. I did a small test, with 10
>> packets only. The result is composed by 18 flows. My output was:
>>
>> # ra -r ARGUS-FILE.argus -M rmon -s stime ltime saddr daddr spkts
>> sbytes sappbytes
>>
>>   18:35:01.691172    18:35:01.691318     13.21.18.5    13.21.21.16
>>   2         3028            0
>>   18:35:01.691172    18:35:01.691318    13.21.21.16     13.21.18.5
>>    0            0            0
>>   18:35:01.691174    18:35:01.691174    13.21.18.25       20.9.2.23
>>     1         1514            0
>>   18:35:01.691174    18:35:01.691174       20.9.2.23    13.21.18.25
>>     0            0            0
>>   18:35:01.691176    18:35:01.691176      13.21.18.8    13.21.23.23
>>    1         1514            0
>>   18:35:01.691176    18:35:01.691176    13.21.23.23      13.21.18.8
>>     0            0            0
>>   18:35:01.691177    18:35:01.691177      13.21.3.17      13.21.18.4
>>      1           60            0
>>   18:35:01.691177    18:35:01.691177      13.21.18.4      13.21.3.17
>>      0            0            0
>>   18:35:01.691178    18:35:01.691178       1.3.7.20     13.21.18.9
>>    1           69            0
>>   18:35:01.691178    18:35:01.691178     13.21.18.9      1.3.7.20
>>   0            0            0
>>   18:35:01.691179    18:35:01.691179       21.1.1.3    13.21.18.19
>>    1         1462            0
>>   18:35:01.691179    18:35:01.691179    13.21.18.19      21.1.1.3
>>   0            0            0
>>   18:35:01.691320    18:35:01.691320    13.21.19.11     13.4.17.20
>>   1           60            0
>>   18:35:01.691320    18:35:01.691320     13.4.17.20    13.21.19.11
>>   0            0            0
>>   18:35:01.691321    18:35:01.691321    13.21.18.13    13.21.23.16
>>    1         1514            0
>>   18:35:01.691321    18:35:01.691321    13.21.23.16    13.21.18.13
>>    0            0            0
>>   18:35:01.691322    18:35:01.691322    13.21.19.14   21.16.10.19
>>   1         1514            0
>>   18:35:01.691322    18:35:01.691322    21.16.10.19    13.21.19.14
>>    0            0            0
>>
>> Is my ra command wrong? Because, as you can see, my sappbytes column
>> is all 0 (zeros).
>>
>> Thanks and regards,
>> Ricardo.
>>
>>
>>
>> On Wed, Oct 26, 2011 at 2:06 PM, Carter Bullard <carter at qosient.com> wrote:
>>> Hey Ricardo,
>>> The idea was that for headers that were of constant size, the data user could subtract the constant from the total bytes to get the answer. For ethernet its, what, 14 bytes per packet.  With VLAN headers its 16, PPP i can never remember, but sems like it 4? IPv4 GRE is, what, 8. But they are constant.
>>>
>>> But for those that are variable length, this doesn't work, so what to do? Didn't seem reasonable to report the sizes of every encapsulation, as that adds a lot of data to the argus record.
>>>
>>> We decided to provide 2 sets of byte counts, total and transport payload, the "app" bytes.  This is what network engineers would use to calculate efficency, (app bytes / total bytes), and when the user bytes don't include retransmissions, the successful app bytes is the metric to use to calculate "goodput".
>>>
>>> That is our rationale.  You can derive metrics like this:
>>>
>>>   ethernetPayload = bytes - (pkts * 14);
>>>   totalHeaderBytes = bytes - appbytes;
>>>   networkCost = ( bytes - appbytes ) / bytes;
>>>   throughPut =  bytes / dur;  ( load )
>>>   goodPut = appbytes / dur;
>>>
>>> What are you going to do with your number?
>>>
>>> Carter
>>>
>>> On Oct 26, 2011, at 4:42 AM, Ricardo S <super.ismiti at gmail.com> wrote:
>>>
>>>> Hello all,
>>>>
>>>> I have a simple question, but reading the manuals I couldn't figure
>>>> out how to solve it. On summing the bytes of a flow, Argus considers
>>>> the Ethernet header, right? If so, how could I remove Ethernet header
>>>> from the total of bytes? Is there any filter expression that would do
>>>> it? I would like to have only the sum of IP headers in the field
>>>> "bytes".
>>>>
>>>> Thanks,
>>>> Ricardo.
>>>>
>>>
>>
>
>



More information about the argus mailing list