In both cases, it is a bandwidth issue. You either consume the bandwidth in the CPU or consume it on the wire. CPU is cheaper and often under-utilized. Of the two, CPU bandwidth is also cheaper/easier to upgrade.
-----Original Message----- From: Rowland, Alan D [mailto:alan_r1@corp.earthlink.net] Sent: Friday, April 27, 2001 8:45 AM To: 'Kurt Kayser'; Tony Hain Cc: Roeland Meyer; John Fraizer; Paul Lantinga; nanog@merit.edu Subject: RE: jumbo frames
I am not an EE but maybe if you rephrased the question as
Which is greater, the cpu cycles to assemble/dissemble jumbo frames or the additional cycles/bandwidth of more numerous ACK packets?
Then again, I may be way out of my depth here.
-Al
-----Original Message----- From: Kurt Kayser [mailto:kurt@noris.de] Sent: Friday, April 27, 2001 8:07 AM To: Tony Hain Cc: Roeland Meyer; John Fraizer; Paul Lantinga; nanog@merit.edu Subject: Re: jumbo frames
Hi,
Isn't it a lot more cpu-intensive to 'collect' some 1500-byte frames into some larger bucket, reassemble it into a jumbo-frame when the next box has to chop it in order to send it out on a Sonet/PPP/etc interface which might have a smaller MTU again?
Doesn't make too much sense to me. I guess that was Tony's aim as well..
Kurt
Roeland you are talking about jumbo frames from the end system lan, while John is talking about only using the jumbo frames between the routers. My point was that in John's environment the packets will all be 1500 since the packets are restricted to that size just to get to the router with the GE interface. I understand that there are perf gains as long as the entire path supports the larger packets, but I don't understand the claim that having a bigger pipe in the middle helps.
Tony
-- noris network AG * tel +49 911 93 52-0 * internet Kilianstraße 142 * fax +49 911 93 52-100 * solution 90425 Nürnberg * http://www.noris.net * provider