Mailing List Archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tlug] Suggestions for poor FTTH performance



I have a stand-alone house so they pulled the fiber all the way into the room I use as an office. The modem is a GE-PON from NTT. I have a RouterBoard router set-up to make the PPPoE connection. As recommended in a number of blogs, I have the MTU on the PPPoE interface set at 1454 (apparently there's some issue with Flet's, although that may or may not be the case with "Flet's Next"). I haven't changed the MTU on the raw connection to the modem but I'm seeing zero error packets or fragmented packets so I assume that's not my problem. Except for an extra NAT rule to the PPPoE interface, my firewall rules are unchanged from before.

Dropping back to the cable-TV connection, I saw over 90Mb early this morning so I'm sure the rest of the network is fine (changing is simply a matter of enabling the cable-TV interface because it's routing "distance" is 1 as opposed to the cable-TV's 0 so once the cable connection is enabled, all new connections route there. I only mention that as a way of eliminating any potential firewall or LAN-internal issues.

I was suspecting throttling and I've written them about it (no reply yet), but... like you say... that much congestion/throttling shouldn't be happening. And it doesn't explain why Yahoo HK (which my wife uses a LOT) won't connect at all -- unless there's also some site-based filtering happening as well. There are other experiments I can do to try to figure out a pattern but I'm kinda pressed for time right now and was just hoping someone might have solved the same problem at some point. But if it's an issue with my local PPPoE setup, I may have to just duke it out on my own. Not very much setup-related information online as far as RouterBoard set-up with Flet's.


On Fri, Apr 14, 2017 at 11:57 AM, Curt Sampson <cjs@example.com> wrote:
As Furkan points out, there are two different setups that are common:
fiber direct from the Central Office to your rooms, and (usually
shared) fiber to the MDF (Main Distribution Frame) in an apartment
building and then SDSL over copper (usually the in-building telephone
lines) from the MDF to your apartment. The former I believe is usually
called "home fiber" and the latter "mansion fiber."

You can usually tell the former because you have actual optical fiber
in one of your rooms, connected to a modem that turns it into
Ethernet. For the latter, you'll have copper (usually from your
telephone jack) coming into the SDSL modem which is converted to
Ethernet on the other side.

When I got my Mansion fiber I watched the installer test the SDSL and
he was getting somewhere betweeen 80 and 90 Mbps. I think I actually
got a document then or later documenting this test as well. You'll
note that this is a lot faster than an ADSL link all the way back to
the CO; this is as expected since it's a much shorter link. (Typically
a few dozen or a hundred meters rather than hundreds or even thousands
of meters.)

If you're on Mansion fiber especially, and you have any number of other
folks in your building, your speeds may be heavily affected by the time
of day. That's one of the first things I'd check out.

Anyway, whichever sort of local loop you end up with, eventually your
link goes back to the CO where it gets hooked up through NTT's
metropolitan network to your ISP, and everything beyond that point is
on the ISP, not the local loop provider. And that's usually where the
bandwidth restrictions happen.

I use Asahi as one of my two Internet providers at home (both go over
the same Mansion fiber). They definitely do throttling, because I've
seen dramatic examples of this on massive outbound web server traffic
(which is not unexpected, this was a dramatic misuse of a home
Internet link). Personally I seem to get about the full bandwidth of
my connection (around 9 MB/sec) when downloading games from Steam for
the first 20 GB or so, and then the rate drops to 1 MB/sec. Pausing
the download and letting it sit for 15-30 minutes seems to let it
restart at 9 MB/sec. again. This could be related to how Steam provides
the downloads, too, but Asahi seems to be the more likely culprit.

Given that this appears to be way beyond the performance you're getting,
my guess would be that it's not likely Asahi that's throttling you; it's
either problems with your local loop or the testing sites you're using.
A quick check with testmy.net at work gave me a 49.3 Mbps download via
gigabit fiber and OCN as the ISP, so these guys don't seem too bad, but
it still seemed that they were sending from a server in Texas, which is
not how you want to be testing. (You're more likely to see throttling of
various sorts on cross-ocean links than to local content providers.) I
suggest you start by checking with sites that have content distribution
from servers in Japan. Using the `mtr` program can help with this.

cjs
--
Curt Sampson         <cjs@example.com>         +81 90 7737 2974

To iterate is human, to recurse divine.
    - L Peter Deutsch

--
To unsubscribe from this mailing list,
please see the instructions at http://lists.tlug.jp/list.html

The TLUG mailing list is hosted by ASAHI Net, provider of mobile and
fixed broadband Internet services to individuals and corporations.
Visit ASAHI Net's English-language Web page: http://asahi-net.jp/en/


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links