RE: Strange public traceroutes return private RFC1918 addresses
And why 4470 for POS? Did everyone borrow a vendor's FDDI-like default or is there a technical reason? PPP seems able to use 64k packets (as can the frame-based version of GFP, incidentally, POS's likely replacement).
According to this URL http://www.columbia.edu/acis/networks/advanced/jumbo/jumbo.html which you have seen before, the number of CRC bits in the protocol header limits the number of bytes you can practically use for the MTU. I expect that we won't go beyond 9000 byte MTUs for a long time. The 4470 for POS probably comes from Token Ring originally. In the original 4 Mbps token ring a device was allowed to hold the token for 9.1 ms which was enough time to transmit 4550 octets. This timing was probably adopted by FDDI which borrowed a lot from the token ring design. No doubt, the designers of POS were also influenced by token ring and just chose the same size. --Michael Dillon
How does a 50Mbyte MTU sound like? http://www.psc.edu/~mathis/MTU/ ~Hani Mustafa
Hani Mustafa <hani.mustafa@noorgroup.net> writes:
How does a 50Mbyte MTU sound like?
Sounds like a 10 terabyte packet buffer per interface in the router to me. Sure will be spiffy to have a nice fine granularity of 50 megs per packet for TCP backoff - handy thing that ever since rfc1323 you can have 32-bit window sizes, or were these guys proposing dispensing with the window altogether since 99.9999% of the time the data you have to send will be under 50 megs? I'll channel Randy for a moment and encourage my competitors to do this... on their *own* networks, please. ---Rob
On 4-feb-04, at 23:46, Hani Mustafa wrote:
How does a 50Mbyte MTU sound like?
Excessive. The trouble is that you get to throw away much more data for a single bit error and also that the difference between the average packet size and the maximum would be huge, so any statically allocated buffers would essentially be a waste of memory most of the time.
Ok, I know that this is getting away from the original thread, but I've always wondered this... Why is the MTU on Ethernet 1500 bytes? I have looked through various docs (eg IEEE Std 802.x) and can find where maxUntaggedFrameSize is listed as 1518 octets, but there is no mention of why this was chosen. I know where the minimum frame size comes from (CSMA/CD and propagation times, etc), but the maximum frame size number sounds fairly arbitrary. -- Warren. On Feb 4, 2004, at 5:46 PM, Hani Mustafa wrote:
How does a 50Mbyte MTU sound like?
http://www.psc.edu/~mathis/MTU/
~Hani Mustafa
"Build a man a fire, and he'll be warm for a day. Set a man on fire, and he'll be warm for the rest of his life." -- Terry Pratchett
Warren Kumari wrote:
Ok, I know that this is getting away from the original thread, but I've always wondered this...
Why is the MTU on Ethernet 1500 bytes? I have looked through various docs (eg IEEE Std 802.x) and can find where maxUntaggedFrameSize is listed as 1518 octets, but there is no mention of why this was chosen. I know where the minimum frame size comes from (CSMA/CD and propagation times, etc), but the maximum frame size number sounds fairly arbitrary.
Because that was a conveniently large amount of very pricey memory availble at the time? Because that was the amount that could be blatted down a 500 meter hose and get the "CD" part to work at some common clock rate?
participants (6)
-
Hani Mustafa -
Iljitsch van Beijnum -
Laurence F. Sheldon, Jr. -
Michael.Dillon＠radianz.com -
Robert E. Seastrom -
Warren Kumari