+1 to what Joel said. It is clear that Joe's view of "fast enough" for a host interface is not the same as my customer's view of "fast enough" for a core router (nor should it be the same -- routers have to be faster than host interfaces). More important is the issue of what can we get vendors to implement, which is probably better worded as what can we get network operators to push vendors to implement. If you are talking about fundamentally changing the internal data path for packets within a router or changing the data path into and out of specific chips to carry significantly more information, then we are talking about at least hundreds of millions and probably billions of dollars of equipment that needs to be replaced. This needs a very compelling justification before it is actually going to happen. Ross -----Original Message----- From: mpls [mailto:mpls-bounces@xxxxxxxx] On Behalf Of Joel M. Halpern Sent: Monday, January 27, 2014 2:19 PM To: Joe Touch; EXT - joelja@xxxxxxxxx; stbryant@xxxxxxxxx Cc: mpls@xxxxxxxx; IETF discussion list Subject: Re: [mpls] Last Call: <draft-ietf-mpls-in-udp-04.txt> (Encapsulating MPLS in UDP) to Proposed Standard Yes Joe, routers could ahve been built to do those calcualtions at that performance scale. There are however two major problems: 1) That is not how routers are built. 2) The target performance scale is rather higher. So could someone build an ASIC to do what you want? Probably. Is there any reason in the world to expect operators to pay the significant extra cost for such? Not that I can see. And even if we could and they would, that is not the world into which we are deploying these tunnels. Yours, Joel On 1/27/14 1:53 PM, Joe Touch wrote: > > > On 1/27/2014 10:48 AM, joel jaeggli wrote: >> On 1/27/14, 8:48 AM, Joe Touch wrote: >>> Those same mechanisms have provided hardware checksum support for a >>> very long time. >> >> The new header and the payload are actually in different parts of the >> forwarding complex until they hit the output queue, you can't checksum >> data you don't have. > > You can (and some do) the checksum component parts when things go into > memory; the partial sums can be added as the parts are combined in the > output queue. > > I appreciate that we're all taking about what might be done, but the > reality is that there are many 'transparent TCP proxies' that have to do > this, so there's clearly a solution, and it clearly runs fast enough. > > Joe > _______________________________________________ > mpls mailing list > mpls@xxxxxxxx > https://www.ietf.org/mailman/listinfo/mpls > _______________________________________________ mpls mailing list mpls@xxxxxxxx https://www.ietf.org/mailman/listinfo/mpls