The origin of this discussion is zero UDP checksums and http://www.ietf.org/mail-archive/web/ietf/current/msg85354.html I suggest reading Jonathan Stone's papers and PhD thesis for a good understanding of where errors can come from. Lloyd Wood http://about.me/lloydwood ________________________________________ From: mpls [mpls-bounces@xxxxxxxx] On Behalf Of Curtis Villamizar [curtis@xxxxxxxxxxxxxx] Sent: 10 January 2014 02:17 To: David Allan I Cc: mpls@xxxxxxxx; tsvwg@xxxxxxxx; lisp@xxxxxxxx; ietf@xxxxxxxx Subject: Re: [mpls] misdelivered mpls packets - Was: Re: draft-ietf-mpls-in-udp was RE: gre-in-udp draft Since we are all top posting ... For counting dropped packets LM OAM is what you need. CV doesn't do that. Direct LM will give best results on a per LSP basis but requires forwarding chip support to do so. I also don't know the origin of this discussion. >> I am interested in how often in practice MPLS packets get >> misdelivered due to label corruption. Not very often if ever at least for major vendor products. I don't know if second tier or third tier players have bugs. (Very often circa 2000 or earlier particularly with one specific vendor and with one provider who made things much worse by mucking with the ILM through management plane, but that is ancient history). Bit rot in TCAM or SRAM is very rare. Bit rot in circa 1980s DRAM was a problem. Bit rot in modern non-ECC DRAM is rare. Bit rot in modern ECC DRAM is very rare. Therefore to have a bad ILM you need bugs. Curtis In message <E6C17D2345AC7A45B7D054D407AA205C391FEA0A@xxxxxxxxxxxxxxxxxxxxxx> David Allan I writes: Hi (trimmed received list) I'm not sure what the starting point of this dialog is, but I can at least say Greg is right on the CV front, it only detects and does not measure misdirection. The only other possibility to count anything is a "no ILM" condition where the label value is unrecognized by the receiving LSR, which IMO could not be made authoritative if it is label values in either the frame or the NHLFE and not exclusively next hops that is corrupted. Cheers D -----Original Message----- From: mpls [mailto:mpls-bounces@xxxxxxxx] On Behalf Of Gregory Mirsky Sent: Thursday, January 09, 2014 9:30 AM To: Loa Andersson; mark.tinka@xxxxxxxxx; stbryant@xxxxxxxxx Cc: gorry@xxxxxxxxxxxxxx; mpls@xxxxxxxx; lisp@xxxxxxxx; ietf@xxxxxxxx; david.black@xxxxxxx; jnc@xxxxxxx; tsvwg@xxxxxxxx Subject: Re: [mpls] misdelivered mpls packets - Was: Re: draft-ietf-mpls-in-udp was RE: gre-in-udp draft Hi Loa, et al., I think that you're referring to Connectivity Verification in MPLS-TP (RFC 6428). It would only detect mis-connection but would not count leaked in frames. Such problem may be more apparent in Segment Routing (SPRING WG). Perhaps CV should be a requirement in SR OAM. Regards, Greg -----Original Message----- From: mpls [mailto:mpls-bounces@xxxxxxxx] On Behalf Of Loa Andersson Sent: Thursday, January 09, 2014 8:17 AM To: mark.tinka@xxxxxxxxx; stbryant@xxxxxxxxx Cc: gorry@xxxxxxxxxxxxxx; mpls@xxxxxxxx; ietf@xxxxxxxx; david.black@xxxxxxx; tsvwg@xxxxxxxx; jnc@xxxxxxx; lisp@xxxxxxxx Subject: [mpls] misdelivered mpls packets - Was: Re: draft-ietf-mpls-in-udp was RE: gre-in-udp draft Changed the subject line ! On 2014-01-09 20:36, Mark Tinka wrote: > On Thursday, January 09, 2014 02:08:43 PM Stewart Bryant > wrote: > >> Either or both. > > I can only speak to native MPLS, as I've never run tunneled MPLS. > >> I am interested in how often in practice MPLS packets get >> misdelivered due to label corruption. > > Well, first of all, routers would need to report corrupted MPLS frames > so operators can glean this data. This isn't something I've come > across, but it would be good to find some kind of way to count this > across interfaces, if the routers can detect and report them. This would be possible to do with MPLS-TP OAM, wouldn't it? /Loa > > The known issue about mis-delivery of MPLS frames is poorly- sized MTU > interfaces. I have no empirical data as to how this can corrupt > successive MPLS frames that may fit into the transit MTU. But in this > case, as with any Layer 2 traffic, not enough MTU = dropped frame. > > Mark. > -- Loa Andersson email: loa@xxxxxxxxxxxxxxxxx Senior MPLS Expert loa@xxxxx Huawei Technologies (consultant) phone: +46 739 81 21 64 _______________________________________________ mpls mailing list mpls@xxxxxxxx https://www.ietf.org/mailman/listinfo/mpls _______________________________________________ mpls mailing list mpls@xxxxxxxx https://www.ietf.org/mailman/listinfo/mpls _______________________________________________ mpls mailing list mpls@xxxxxxxx https://www.ietf.org/mailman/listinfo/mpls _______________________________________________ mpls mailing list mpls@xxxxxxxx https://www.ietf.org/mailman/listinfo/mpls