Hi Tom, agree, single timestamp may be used as a marker too. But without addition of a sequence number accuracy of your measurement will be affected by packet loss, re-ordering and packet duplication. And I don't think that 32 bits-long timestamp format enables high precision of delay/delay variation calculation. AFAIK, delay measurement protocols use 64 bits-long formats, NTP or PTPv1/truncated PTPv2. RTT would work only for bi-directional applications, though it simplifies operations as it doesn't require clock synchronization in the network.
Regards, Greg On Tue, Aug 9, 2016 at 10:56 AM, Tom Herbert <[email protected]> wrote: > On Tue, Aug 9, 2016 at 9:56 AM, Greg Mirsky <[email protected]> wrote: > > Hi Tom, > > I've read a proposal to apply timestamps to data packets on a fly, I > believe > > submitted in the SFC WG. I cannot characterize such method as passive as, > > per my understanding, it will increase the size of the packet to > accommodate > > the timestamps. The advantage of the Alternate Marking method is that > once > > marker applied the measurements of packet loss and delay/delay variation > > could be performed at numerous test points along the path or even within > a > > node without any changes to the packet itself. To achieve the same with > > direct timestamping, I think, would require collection of N+1 64 > bits-long > > timestamps. > > > I was thinking more that only the sender would place a time stamp in > the packet. Any device on the path that sees the timestamp can > determine the one way latency from the sender as the difference of > time packet is received and the timestamp assuming the clocks are > synchronized. The timestamp allows measurements to be taken for each > packet at the cost of overhead of timestamp in packet which could be > as little four bytes. Alternatively, the timestamp could also be used > to measure RTT if it is reflected by the destination end point (like > how TCP timestamp is used) > > Tom > > > > Regards, Greg > > > > On Tue, Aug 9, 2016 at 9:28 AM, Tom Herbert <[email protected]> wrote: > >> > >> On Tue, Aug 9, 2016 at 9:07 AM, Greg Mirsky <[email protected]> > wrote: > >> > Hi Tom, > >> > many thanks for the most informative response. I've added mu notes > >> > in-line > >> > under tag GIM>>. > >> > > >> > Regards, Greg > >> > > >> > On Mon, Aug 8, 2016 at 5:44 PM, Tom Herbert <[email protected]> > wrote: > >> >> > >> >> On Sun, Aug 7, 2016 at 7:02 PM, Greg Mirsky <[email protected]> > >> >> wrote: > >> >> > Dear Authors of the VxLAN-GPE, GUE, and GENEVE, > >> >> > all protocols under consideration use a bit flag rather than > explicit > >> >> > protocol type to indicate that payload is a test packet, i.e. > active > >> >> > OAM. > >> >> > I'm trying to understand the rationale of such decision. Does use > of > >> >> > the > >> >> > bit > >> >> > flag rather than protocol type produce more efficient > implementation, > >> >> > is > >> >> > more HW friendly? In GUE, the the field to indicate type of the > >> >> > payload > >> >> > even > >> >> > tagged Proto/ctype as its interpretation depends upon value of the > C > >> >> > bit. > >> >> > >> >> The C-bit in GUE distinguishes data messages from control > >> >> messages.Data messages are considered to be the payload of > >> >> encapsulation, whereas control messages are about the encapsulation > >> >> itself. OAM might be one type of control message in GUE, however > there > >> >> could be others. For instance if we wanted some sort of negotiation > >> >> between two endpoints to exchange capabilities or supported features > >> >> this would fit well into a control message. > >> > > >> > GIM>> Yes, what I've proposed is clearly more than just OAM channel. > In > >> > fact, it is Associated Channel (ACh) that may be used by control, > >> > management > >> > and OAM. And as I've used term "Associated Channel" you'll easily > >> > recognize > >> > that I have MPLS background and draw on MPLS/MPLS-TP OAM experience. > And > >> > as > >> > Generic ACh (G-ACh) is used to advertise capabilities of an LSR in RFC > >> > 7212, > >> > AC-h in NVO3 can support similar functionalities as well. > >> >> > >> >> > >> >> > But wouldn't it be simpler if all proposals used protocol type to > >> >> > identify > >> >> > OAM payload? And if the protocol type is OAM, then after the > protocol > >> >> > header > >> >> > have OOAM Header, e.g. as proposed in . Then > >> >> > >> >> Each of the three protocols has a protocol next header field, however > >> >> the field is defined differently among them. The next header in GUE > is > >> >> an IP protocol number, in Geneve it is an Ethertype, and VXLAN-GPE > >> >> uses a new number space. In GUE we could probably use ICMP protocol > >> >> for OAM by defining the appropriate types (that might have the > >> >> advantage of allow OAM to be generic instead of restricted to only > >> >> encapsulation). Presumably, VXLAN-GPE could define some value in the > >> >> number space for for OAM. For Geneve maybe there is an appropriate > >> >> Ethertype? > >> >> > >> >> > NVO3 protocols would be able to have common Active OAM (Fault > >> >> > Management > >> >> > and > >> >> > Performance Measurement) that can be used in BIER and SFC. And the > >> >> > bit, > >> >> > the > >> >> > bit I'd propose to redefine to be used for passive performance > >> >> > measurement > >> >> > as described in draft-ietf-bier-pmmm-oam. (Allocating two bits-long > >> >> > field > >> >> > would enable more accurate measurements using the Alternate Marking > >> >> > method). > >> >> > And these steps will enable us to develop common Active OAM and use > >> >> > passive > >> >> > performance measurement regardless, almost, of the data plane > >> >> > protocol > >> >> > used > >> >> > in NVO layer. > >> >> > >> >> The problem I see with trying to constrain the solution to only one > or > >> >> two bits of information is that this substantially limits the > >> >> functionality. With an extensible protocol we should be able more > >> >> information to get more accurate measurement. For instance, I might > >> >> want to measure the latency of individual packets to get feedback on > >> >> path selection, correlate performance to packet loss, etc. Has the > OAM > >> >> DT considered the requirements and solutions for passive performance > >> >> measurement? > >> > > >> > GIM>> Indeed, the OAM DT had considered the requirements to enable use > >> > of > >> > performance measurement methods as passive OAM. Should note that we > use > >> > term > >> > "passive method" somewhat differently from the definition in RFC 7799. > >> > Such > >> > interpretation was discussed in the IPPM WG and we've agreed that if a > >> > measurement method does not change treatment of a data packet by the > >> > network > >> > (e.g. doesn't change its CoS, length or else), then the method behaves > >> > as > >> > close as passive and may be characterized as such. Measurements for a > >> > single > >> > packet are possible using the Alternate Marking method with two > >> > bits-long > >> > marker. The draft in BIER WG has such example. I've attached the > >> > presentation slides. Will be glad to answer any further questions. > >> > >> Yes, but number of specific packets I could measure is still limited > >> in some time quantum with the two bit method. Alternatively, if every > >> packet contained a timestamp for instance, then we can measure every > >> packet and filter or aggregate the measurements with arbitrary > >> granularity of our choosing. BIER may have been defined with as a > >> fixed length header so that a couple of bits are all that could > >> feasibly be allocated to OAM, but this is not necessarily true for > >> other encapsulation protocols that are purposely extensible to support > >> a richer set of features. > >> > >> Tom > >> > >> >> > >> >> > >> >> Thanks, > >> >> Tom > >> >> > >> >> > > >> >> > Regards, Greg > >> >> > > >> >> > On Fri, Jul 29, 2016 at 8:13 AM, Alia Atlas <[email protected]> > >> >> > wrote: > >> >> >> > >> >> >> I'd like to have people focus on the key point of this thread. > >> >> >> > >> >> >> Are there serious technical objections (and specifically what are > >> >> >> they) > >> >> >> to > >> >> >> moving forward with VXLAN-GPE as the standards-track protocol? > >> >> >> > >> >> >> Are there serious technical objections (and specifically what are > >> >> >> they) > >> >> >> to > >> >> >> moving forward with GENEVE as the standards-track protocol? > >> >> >> > >> >> >> Are there serious technical objections (and specifically what are > >> >> >> they) > >> >> >> to > >> >> >> moving forward with GUE as the standards-track protocol? > >> >> >> > >> >> >> We need to capture any relevant objections. So far, there's been > >> >> >> some > >> >> >> discussion on extensibility - with Tom Herbert providing concrete > >> >> >> concerns. > >> >> >> > >> >> >> I have concluded that almost all the authors would prefer to have > no > >> >> >> standards track solution if they can't guarantee that theirs is > that > >> >> >> standard. > >> >> >> > >> >> >> I do hear concerns about whether a decision will be too late. I > >> >> >> think > >> >> >> that a decision can only be helpful. It goes back to when is the > >> >> >> best > >> >> >> time > >> >> >> to plant a tree - with the answer of 20 years ago or now. > >> >> >> > >> >> >> Regards, > >> >> >> Alia > >> >> >> > >> >> >> > >> >> >> On Fri, Jul 29, 2016 at 4:34 AM, Naoki Matsuhira > >> >> >> <[email protected]> wrote: > >> >> >>> > >> >> >>> > >> >> >>> > >> >> >>> On 2016/07/21 23:56, Bocci, Matthew (Nokia - GB) wrote: > >> >> >>>> > >> >> >>>> WG > >> >> >>>> > >> >> >>>> There was a discussion in the NVO3 WG meeting in Berlin > following > >> >> >>>> strong > >> >> >>>> advice from our Area Director that we need to come to a > consensus > >> >> >>>> on > >> >> >>>> converging on a common encapsulation. Two sets of questions were > >> >> >>>> asked: > >> >> >>>> (1) Should the WG move forward with one standards track encap? > >> >> >>>> (2) For a given encap, do you have significant technical > >> >> >>>> objections? > >> >> >>> > >> >> >>> > >> >> >>> I want to inform to this mailing list that I proposed ME6E-FP and > >> >> >>> ME6E-PR > >> >> >>> at the yokohama meeting. I also have proposal M46E-FP and M46E-PR > >> >> >>> (past > >> >> >>> called SA46T). > >> >> >>> > >> >> >>> These encapsulation technologies are based on address mapping. > ME6E > >> >> >>> use > >> >> >>> IPv6 address which mapping MAC address, and M46E use IPv6 address > >> >> >>> which > >> >> >>> mapping IPv4 address. > >> >> >>> > >> >> >>> I understand too many encapsulation technologies, however these > my > >> >> >>> proposal are simple, and may contribute to the Internet. > >> >> >>> > >> >> >>> I believe address mapping approach is unique, so I want to > propose > >> >> >>> again. > >> >> >>> > >> >> >>> sorry not the answer to the question. > >> >> >>> > >> >> >>> Naoki Matsuhira > >> >> >>> > >> >> >>> > >> >> >>> _______________________________________________ > >> >> >>> nvo3 mailing list > >> >> >>> [email protected] > >> >> >>> https://www.ietf.org/mailman/listinfo/nvo3 > >> >> >> > >> >> >> > >> >> >> > >> >> >> _______________________________________________ > >> >> >> nvo3 mailing list > >> >> >> [email protected] > >> >> >> https://www.ietf.org/mailman/listinfo/nvo3 > >> >> >> > >> >> > > >> >> > > >> >> > _______________________________________________ > >> >> > nvo3 mailing list > >> >> > [email protected] > >> >> > https://www.ietf.org/mailman/listinfo/nvo3 > >> >> > > >> > > >> > > > > > >
_______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
