XGS-PON is one of the >> 1Gbps successor technologies for GPON. GPON typically is asymmetric with ~2.4/1.2 Gbps "segment capacity" (but as far as I understand the ITU standard also allows symmetric 2.4/2.4 GPON). XG-PON ups the "ceiling" to a marketing-friendly 10Gbps, and XGS-PON makes that symmetric, for 10/10, and it does so fully backward compatible with GPON (by using different wavelength) so the same PON can have (different) users with XGS-PON and GPON, albeit the CPE are currently either GPON or XGS-PON and the at the head end one obviously needs two different illumination sources as well. As far as I can tell both use a request-grant mechanism to coordinate the CPE in egress direction (towards the OLT) but the cycle time is < 1ms IIRC. ITU standards documents are no joy to read (and everything appears to my naive mind quite baroque), so I probably am overlooking things and grossly simplifying others.... Ad an end-user I would welcome a switch from DOCSIS to XGSPON (not that I use either, still on 100/40 VDSL2, but I could switch to 1000/50 DOCSIS if I wanted*), but I am sure the devil is in the details and it will be possible to spoil true fiber as well. I note that in France and Switzerland**, ISPs already offer XGS-PON links to end-users... in a PON of up to 32 CPE, they offer each user pans up to 10Gbps, they clearly assume that no one is going to either saturate that long enough to be noticeable or that most people will not care (given how hard it is going tp be tp saturate that with normal use cases). Given that pool of existing users maybe we can find some to run flent tests?
*) But I do not want to, I am not convinced my potentialcable ISPs has truly thought about how to operate its segments well "at capacity" but ever since rolling-up the local market with cheap gigabit (1000/50 for ~40EUR/month) cable segments are prone to overload during primetime. **) In Switzerland one ISP (Init7) even offers symmetric 25/25 Gbps (Fiber7-X2: 25G SFP28 BIDI LR, TX1270/RX1330 nm) as active optical network (25GBASE-LR ethernet), for 777/12 = 64.75 SFR/month (plus 333SFR set-up cost covering the upstream optics and additionally the own CPE). > On Nov 9, 2021, at 13:57, Dave Taht <[email protected]> wrote: > > I know there is one form of PON that has enormous (and unfixible) > potential for bufferbloat, another that doesn't, and other tech like > mpls that can also muddy the issue. Does anyone "get", > comprehensively, where the fiber market is going? > > https://community.virginmedia.com/t5/Speed/Is-DOCSIS-PIE-enabled-per-specification-on-VM/m-p/4861693#M281139 > > -- > I tried to build a better future, a few times: > https://wayforward.archive.org/?site=https%3A%2F%2Fwww.icei.org > > Dave Täht CEO, TekLibre, LLC > _______________________________________________ > Bloat mailing list > [email protected] > https://lists.bufferbloat.net/listinfo/bloat _______________________________________________ Bloat mailing list [email protected] https://lists.bufferbloat.net/listinfo/bloat
