Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: stds-80220-requirements: 802.1q/p





Dave,

Thanks for the clarification.  PPPoE certainly adds overhead relative to
ethernet.  But it's also possible to use PPP over the air without PPPoE.
3GPP2, for example, directly transmits PPP encapsulated IP over its air
interface (i.e., with no ethernet or PPPoE headers).  By using synchronous
(RFC 1662) PPP, the overhead can be reduced to two bytes per IP packet as I
described in my earlier message.

The bottom line here is, in my opinion, the following.  The requirements
and design considerations for the air interface are essentially the same
whether it's meant to bridge IP-in-PPP, IP-in-ethernet, IP-in-PPPoE.  The
fundamental design problem in all cases is the following: design an air
interface to efficiently convey IP traffic encapsulated in a thin protocol.
While there are differences in the detailed handling for each
encapsulation, e.g., for the extraction of IP layer QoS markings for use by
the radio resource scheduler, none of them are significant enough at this
point to either rule out or to mandate the use of any particular
encapsulation.

Marc

------- start of forwarded message (RFC 934 encapsulation) -------
From: "Mcginniss, Dave S [GMG]" <david.s.mcginniss@mail.sprint.com>
To: "Marc Goldburg" <marcg@arraycomm.com>
Cc: "Branislav Meandzija" <bran@arraycomm.com>,
   <stds-80220-requirements@ieee.org>
Subject: RE: stds-80220-requirements: 802.1q/p
Date: Fri, 21 Nov 2003 16:27:29 -0600

I am assuming PPPOE

5) Analysis of PPPoE overhead 


   From the PPPoE Informational RFC document, we see that: 


      The Maximum-Receive-Unit (MRU) option MUST NOT be negotiated to a 
      larger size than 1492. Since Ethernet has a maximum payload size 
      of 1500 octets, the PPPoE header is 6 octets and the PPP Protocol
ID 
      is octets, the PPP MTU MUST NOT be greater than 1492. 


   This means that PPPoE imposes an overhead of 8 bytes for every 1500 
   bytes of data for as long as the negotiated MTU is 1492 bytes. 8 over

   1500 is thus .53% over overhead on packets that fill up the complete 
   ethernet payload capacity. But if the PPP MTU is negotiated to be 
   smaller, in order to improve the performance of VoIP or real-time 
   gaming applications for example, the overhead becomes drastically
more 
   apparent on smaller packets. For example, the MTU on dial-up modems 
   is usually as small as 256 bytes. With a 256 bytes MTU, PPPoE would 
   thus present an overhead of 3.125%. This number may not be small, but

   this means that a voice packet, would incur a total rountrip delay 
   that is over 6.25% higher than it would be without the overhead of 
   PPPoE. 



David S. McGinniss
Sprint Broadband Wireless Group
Principal Engineer II 
(630) 926-3184
david.s.mcginniss@mail.sprint.com
 
 

- -----Original Message-----
From: Marc Goldburg [mailto:marcg@arraycomm.com] 
Sent: Thursday, November 20, 2003 21:18
To: Mcginniss, Dave S [GMG]
Cc: Branislav Meandzija; stds-80220-requirements@ieee.org
Subject: RE: stds-80220-requirements: 802.1q/p


On Thu, Nov 20, Mcginniss, Dave S [GMG] writes ...

Dave> PPP also increases overhead.

Dave,

Relative to ethernet, I don't believe that this is correct.  PPP adds 2
bytes of overhead to an IP packet.  Standard ethernet adds 14 bytes
(802.3
adds 17), and 802.1Q adds an additional 4 bytes.

Regards,

Marc
 

------- end -------