Date: Mon, 22 Mar 1999 07:37:48 -0500 From: "Louis A. Mamakos" <louie@TransSys.COM> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: tlambert@primenet.com, hasty@rah.star-gate.com, wes@softweyr.com, ckempf@enigami.com, wpaul@skynet.ctr.columbia.edu, freebsd-hackers@FreeBSD.ORG Subject: Re: Gigabit ethernet -- what am I doing wrong? Message-ID: <199903221237.HAA38608@whizzo.transsys.com> In-Reply-To: Your message of "Sun, 21 Mar 1999 23:40:20 PST." <199903220740.XAA16463@apollo.backplane.com> References: <199903220545.WAA10719@usr01.primenet.com> <199903220740.XAA16463@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> :> The solution to this at MAE-WEST was to clamp down on the idiots who > :> were selling transit at MAE-WEST and overcommitting their ports, plus > : > :With respect, technology should operate in the absence of human > :imposition of policy. It should have been technically impossible > :for the idiots to successfully engage in the behaviour in the first > :place, and if it wasn't, then that's a design problem with the > :gigaswitches. > > With respect, you are assuming that the problems can be solved trivially. > These are *NOT* trivial problems. Very not trivial problems. Not even > *CLOSE* to trivial problems. I can't repeat this enough times. Alternatively, you buy hardware that doesn't have pathological performance in certain operating regions. The head-of-line blocking experienced in the DEC FDDI gigaswitches is a classic and well-known characteristic of that type of switch fabric. It has other interesting characteristics, such as having to work hard to avoid deadlocking the switch fabric when multiple input ports are trying to deliver a multicast/broadcast frame to to overlapping output ports. You have to implement partial completion in the the crossbar fabric to avoid this problem. The human intervention that Matt references wasn't tuning the operation of the swtich directly; it was arranging to avoid the part of the operating regime where the swtich exhibits poor performance. If the switch wasn't broken, you wouldn't have to indirectly engineer loading to avoid head of line blocking. But in any case you still have to do large-system type engineering to avoid the 10 pounds of packets in a 5 pound sack problem. You just can't think of these switches in the context of a general purpose computer with busses and peripherals. High-performance switches and routers are fundamentally architecturally different beasts than the the FreeBSD box on your desk. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199903221237.HAA38608>