Ethereal-dev: Re: [Ethereal-dev] Updates to io-stat calculations
Note: This archive is from the project's previous web site, ethereal.com. This list is no longer active.
From: "Ronnie Sahlberg" <ronnie_sahlberg@xxxxxxxxxxxxxx>
Date: Mon, 28 Apr 2003 18:57:27 +1000
Very pretty graphs but there seems to be some semi-serious issues with it. First it seems it only looks at the TCP layer and thus should only be able to produce the graphs reliably iff the client is singlethreaded (only does one command at a time) compared to ehtereal's measurements that are based on data in the actual smb/oncrpc/dcerpc/... layers. But the graphs sure looks very much better than the ethereal ones. It would be very useful if someone hacked up some scripts to take the output from tethereal -z io,stat,0.010,MIN/MAX/AVG(smb.time)smb.time... did some grep and sed magic on it and fed it into gnuplot to generate nice PNGs with smoothed graphs. This would be a very useful thing. ----- Original Message ----- From: "Visser, Martin (Sydney)" Sent: Monday, April 28, 2003 11:31 AM Subject: RE: [Ethereal-dev] Updates to io-stat calculations While not a "sniffer" per-se , Packeteer PacketShaper does quite a nice job of plotting response times etc using histogram buckets. PacketShaper inspects and records stats for all traffic that match "classes", and in this case those that you nominate to record response time. It also does some interesting calculations to work out network time-of-flight and server response (by comparing SYN-ACK response with normal payload response time) Graphically results are output as attached (for telnet traffic on a link to a particular site There is some info on the function at http://support.packeteer.com/documentation/packetguide/current/nav/tasks /rtm/monitor-rtm.htm a And the tech details on RTM calcs at http://support.packeteer.com/documentation/packetguide/current/info/rtm- technical-details.htm Martin Visser Network Consultant Technology & Infrastructure - Consulting & Integration HP Services 3 Richardson Place North Ryde, Sydney NSW 2113, Australia Phone *: +61-2-9022-1670 Mobile *: +61-411-254-513 Fax 7: +61-2-9022-1800 E-mail * : martin.visserAThp.com -----Original Message----- From: Ronnie Sahlberg [mailto:ronnie_sahlberg@xxxxxxxxxxxxxx] Sent: Thursday, 24 April 2003 10:47 PM To: ethereal-dev@xxxxxxxxxxxx Subject: [Ethereal-dev] Updates to io-stat calculations I just checked in some updates to tethereal io-stat calculations. Tethereal can now, in addition ot frames/bytes counts, also calculate COUNT,SUM,MIN,MAX,AVG for several types of fields. Please see manual page for tethereal. Example: tethereal ... -z "io,stat,0.100,ip.addr==1.1.1.1&&smb.time,MIN(smb.time)ip.addr==1.1.1.1& &smb .time,MAX(smb.time)ip.addr==1.1.1.1&&smb.time,AVG(smb.time)ip.addr==1.1. 1.1& &smb.time" This will calculate statistics in 100ms intervals for all smb responses to/from the host at 1.1.1.1. (only response packets have the smb.time field) The output will be presented in 4 columns: Column1: number of frames/bytes for all such response packets. Column2: MINimum response time seen in the interval Column3: MAXimum response time seen in the interval. Column4: AVeraGe response time seen in the interval. The output should be simple to convert with some sed magic into something excel or any other application capable of producing graphs can import. Note that the example above is simplified and may not be useful in real world since some SMB commands will normally have very long response times (i.e. NOTIFY which normally can take minutes/hours to complete) which will poison the data. It may be nessecary to enhance the filter to remove the influence from those calls. Other interesting protocols to plot the response time for like this is probably nfs (rpc.time) and dcerpc.time. Any other sniffer capable of plotting min/max/average response time from a specific client over time? have fun. ronnie sahlberg _______________________________________________ Ethereal-dev mailing list Ethereal-dev@xxxxxxxxxxxx http://www.ethereal.com/mailman/listinfo/ethereal-dev
- References:
- RE: [Ethereal-dev] Updates to io-stat calculations
- From: Visser, Martin (Sydney)
- RE: [Ethereal-dev] Updates to io-stat calculations
- Prev by Date: Re: [Ethereal-dev] Writing Decoders for Ethereal
- Next by Date: [Ethereal-dev] Bug: wrong packet length (in a TCP packet that is within an ICMPv6 redirect packet...)
- Previous by thread: RE: [Ethereal-dev] Updates to io-stat calculations
- Next by thread: RE: [Ethereal-dev] Updates to io-stat calculations
- Index(es):