This is a static archive of our old Q&A Site. Please post any new questions and answers at ask.wireshark.org.

what’s the scenario to use -s snaplen?

0

Hi,

For large amounts of data transfer like video stream, FTP, I don't really need data but I do want to get the full headers of each frame for analysis. The question is, would that cause any potential problem when doing analysis? For example, incorrect statistics, etc...

If we do an IO graph, will the result be the same if you captured a full trace as compared with a trace using -s <snaplen>?

thanks!

asked 20 Aug '14, 02:20

SteveZhou's gravatar image

SteveZhou
191273034
accept rate: 0%

edited 20 Aug '14, 03:23

grahamb's gravatar image

grahamb ♦
19.8k330206


One Answer:

1

Yes, all statistics will be the same, because each frame will still know it's original length. The payload bytes just aren't there to decode, but that's all. So as long as you don't care about payload contents you can capture just the headers and do your analysis.

answered 20 Aug '14, 02:35

Jasper's gravatar image

Jasper ♦♦
23.8k551284
accept rate: 18%

many thanks for quick response!

(20 Aug '14, 02:43) SteveZhou

by the way, what the typical snaplen to specify to make sure all of the headers will be included? Is there a rule of thumb value?

(20 Aug '14, 02:46) SteveZhou

for normal TCP I go for 64 bytes, but if I need upper layer headers like NetBIOS/SMB as well I might choose 128 or 256, depending on how big they are.

(20 Aug '14, 03:05) Jasper ♦♦