Sure, but first some more detail,
is there a particular protocol you are interested in in particular?
SMB or RPC response times?
I often create advance graphs for service response times for both SMB and
RPC (nfs)
Since the graphs are drawn ove each other from the bottom (fifth) to top I
often produce
Advanced iostat for smb.time and rpc.time to plot response times.
Select tick interval==1s pixelspertick==5 and specify the graphs as:
Graph1: field==smb.time filter==smb.time type==AVG (average) and
drawingstyle==LINE
Graph2: field==smb.time filter==smb.time type==MIN and drawingstyle==BAR
or IMPULSE
Graph3: field==smb.time filter==smb.time type==MAX and drawingstyle==BAR
The same config works fine for rpc.time as well.
For client workload/concurrency I usually plot it as
tick interval==0.010 pixelspertick==2 and a single graph
Graph1: field==rpc.time filter==rpc.time type==LOAD drawingstyle==IMPULSE
this produces beautiful graphs of how much concurrency the clients are
generating (see manpage for LOAD)
it shows very clearly how fast the client issues new i/o and how many i/os
the client issues concurrently, how many i/o/s in flight the client keeps at
a time.
----- Original Message -----
From: "Kevin"
Sent: Monday, October 27, 2003 2:30 PM
Subject: [Ethereal-users] IOStats Advanced Examples?
> Would anybody have an example of the use of IOStats with the Advanced
> Units? I have been unsuccessfully playing with this with TCP traces.
>
> Thanks
> Kevin
>
> _______________________________________________
> Ethereal-users mailing list
> Ethereal-users@xxxxxxxxxxxx
> http://www.ethereal.com/mailman/listinfo/ethereal-users