howie@yo.uucp (howie) (08/30/90)
A couple of questions about the use of ping: 1) Because ping uses ICMP packets, we are not getting a true picture of how long it takes for data to be transferred to another machine, True? Is there an easy way to determine the transfer time of data, other than setting up a connection with FTP, and moving some test data? 2) The man page for ping warns against using ping in shell scripts. Is it really that dangerous to use ping automatically, if the number of packets is limited to say, 3? This would be helpful to our users, but we don't want to cause undue congestion. 3) Is there an alternative to the standard BSD ping that we should consider using? thanks for any help, -- howie uw-beaver!ssc-vax!voodoo!howie
roy@phri.nyu.edu (Roy Smith) (08/30/90)
In article <227@voodoo.UUCP> howie@yo.uucp (howie) writes: > The man page for ping warns against using ping in shell scripts. Is it > really that dangerous to use ping automatically, if the number of packets > is limited to say, 3? I have a shell script which is run by cron every hour. It gives 100 pings to each of 4 hosts on our campus-wide LAN, and massages the output before logging it for later perusal. It lets me spot long-term trends in packet lossage on each of two point-to-point links that connect me with the University backbone. I've been doing this for at least a year without any ill effects that I've noticed. My feeling is that the man page warning is unwarranted. As an example of how useful this is, looking near the end of the current log file, I see that something bad happened last night that I need to look into. Wed Aug 29 13:46:34 EDT 1990 100 100 0% 60/87/740 Wed Aug 29 14:46:33 EDT 1990 100 100 0% 60/110/920 Wed Aug 29 15:46:32 EDT 1990 100 99 1% 60/107/980 Wed Aug 29 16:46:31 EDT 1990 100 100 0% 60/87/600 Wed Aug 29 17:46:31 EDT 1990 100 100 0% 60/83/620 Wed Aug 29 18:46:32 EDT 1990 100 94 6% 60/82/520 Wed Aug 29 19:46:32 EDT 1990 100 92 8% 60/90/520 Wed Aug 29 20:46:31 EDT 1990 100 87 13% 60/180/1020 Wed Aug 29 21:46:32 EDT 1990 100 81 19% 60/108/680 Wed Aug 29 22:46:32 EDT 1990 100 88 12% 60/133/800 Wed Aug 29 23:46:32 EDT 1990 100 82 18% 60/91/400 Thu Aug 30 00:46:32 EDT 1990 100 88 12% 60/131/780 Thu Aug 30 01:46:33 EDT 1990 100 89 11% 60/85/400 -- Roy Smith, Public Health Research Institute 455 First Avenue, New York, NY 10016 roy@alanine.phri.nyu.edu -OR- {att,cmcl2,rutgers,hombre}!phri!roy "Arcane? Did you say arcane? It wouldn't be Unix if it wasn't arcane!"
cliff@garnet.berkeley.edu (Cliff Frost) (08/31/90)
In article <227@voodoo.UUCP>, howie@yo.uucp (howie) writes: |> |> Is there an easy way to determine the transfer time of data, other |> than setting up a connection with FTP, and moving some test data? If you want very simple memory to memory tests, you can send data to the TCP (or UDP) discard port. Tom Ferrin's "netout" does this for TCP, it is available for anonymous ftp from jade.berkeley.edu (128.32.136.9). |> 2) The man page for ping warns against using ping in shell scripts. Is it Calculate how much of your user's bandwidth you'll be using. A few pings on an ethernet are obviously trivial. |> 3) Is there an alternative to the standard BSD ping that we should consider |> using? Ping is more of a network management tool than a throughput tool, although it is useful sometimes to give an idea of latency. For network management you want to use SNMP as much as possible. There is an SNMP mailing list. Mail to snmp-request@nisc.nyser.net to join it. Cliff Frost (415) 642-5360 Central Computing Services <cliff@berkeley.edu> University of California CLIFF AT UCBCMSA Berkeley, CA 94720
howie@yo.uucp (howie) (09/01/90)
Thanks to the folks who responded to my questions about ping. Please note that our mailer is messed up. You can email me at: howie@voodoo.uucp (no matter what the "reply to" field indicates) thanks again for the help, -- howie uunet!bcstec!voodoo!howie