First of all, you cannot test this out over a modem. The machine
that is
sending the ping needs to be hooked up also to a fast link (isdn, 10bt,
ect). And
the ping packet size that needs to be sent should be around 10mb, with a
recursion
within there. I've run this through all types of routers, most being sisco, and
since most machines hooked to a fiberoptic link will go through a router
that doubles
as a 10bt hub, the router should not be a factor.
Beware, I've found that if you continue the ping for more than about
5 min
MCI or USWORST will call yer ass up and tell you to knock it off.
=)
At 03:21 PM 8/24/96 -0400, you wrote:
>> This is evident when one flood pings a machine that is hooked up
>> with a 1.5 mbit or faster link to a t1 or greater (ie this will crash the
>> machine, or lag it to a stop if done right)
>
>Well, I dunno about this... just to test this out, I connected to my
>machine using OS/2's TCP/IP via my modem, started an FTP transfer, and
>used a big ping data size (I think around 16K). It obviously slowed up
>my file transfer, but not by as much as you would expect - in theory, it
>should go FTP/ping/FTP/ping/FTP/ping... but it didn't. It did FTP 16
>times, THEN ping (which was okay, since the packet size for FTP was 1K or
>so). In other words, it creatively shuffled to produce a bigger delay on
>ping than it gave on FTP. Seems to me that this is a pretty low-priority
>process, at least on OS/2. Does this mean that the OS/2 TCP/IP is
>better-quality than SunOS', Linux', IRIX', HPUX'... or any others?
>
>Well, pings to T1 host might be sent to the router, which could very
>well be lower-quality than the OS running elsewhere on the network. Or
>it might be that the flood of packets (which are sent no matter what)
>causes some kind of routing impossibility. There's a whole lot of
>possibilities.
>
+-----------------------------------------------------------+
| Ensure that you have read the CircleMUD Mailing List FAQ: |
| http://cspo.queensu.ca/~fletcher/Circle/list_faq.html |
+-----------------------------------------------------------+
This archive was generated by hypermail 2b30 : 12/07/00 PST