No more ipv6 in Singtel home 1G fibre broadband?

Mach3.2

Great Supremacy Member
Joined
Apr 8, 2011
Messages
72,405
Reaction score
2,463
Code:
                                                                                                                                   Packets               Pings
 Host   Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 27.104.128.1                                                                                                                   0.0%    33    2.3   4.4   2.1  25.2   4.0
 2. 202.65.246.6                                                                                                                   0.0%    33    4.8   4.4   2.2  18.1   3.9
 3. 202.65.246.5                                                                                                                   0.0%    33    2.4   2.8   1.8   7.0   1.3
 4. 202.65.246.162                                                                                                                 0.0%    33    3.0   5.1   2.7  46.5   7.6
 5. (waiting for reply)
 6. 216.218.221.42                                                                                                                 0.0%    33    3.8   4.5   3.5   7.3   1.1

mtr output. No issues for me.
Code:
Running mtr  -w -c 10 -i 1 216.218.221.42:

Start: 2020-09-12T17:31:36+0800
HOST: -						      Loss%   Snt   Last   Avg  Best  Wrst StDev
  1.|-- 1.128.104.27.unknown.m1.com.sg    0.0%    10   57.0  10.7   2.3  57.0  16.8
  2.|-- 6.246.65.202.unknown.m1.com.sg    0.0%    10    2.7   2.9   2.3   4.8   0.7
  3.|-- 5.246.65.202.unknown.m1.com.sg    0.0%    10    2.7   2.8   2.4   3.6   0.3
  4.|-- 162.246.65.202.unknown.m1.com.sg  0.0%    10    2.9   2.9   2.6   3.2   0.2
  5.|-- ???                              100.0    10    0.0   0.0   0.0   0.0   0.0
  6.|-- tserv1.sin1.he.net                0.0%    10  178.5 171.9 170.1 178.5   2.5

Mine's kinda high for some unknown reason. :s22:
 

miloaisdino

Senior Member
Joined
Mar 25, 2016
Messages
843
Reaction score
26
M1's seem to have unoptimised route to HE's tunnelbroker server, ping to the server has been hovering around 170ms for the past month or 2.

singtel has been like this for years! try using the hong kong server instead
 

SNAG

Master Member
Joined
Jan 1, 2000
Messages
4,870
Reaction score
100
Or using the test server you specified.


3fW2G8n.jpg

Hmm - so Singtel is supporting native ipv6 now?
 

xiaofan

High Supremacy Member
Joined
Sep 16, 2018
Messages
30,969
Reaction score
8,616
Hmm - so Singtel is supporting native ipv6 now?

No. I need to set up 6rd in the SingTel router to get it work. It is probably that the test server can not differentiate.

Take note I am an ONT user. YMMV for ONR users (original ONR or bridged ONR). Please contact SingTel to confirm.

SingTel "Mesh" router has two IPv6 settings and only 6rd works for me.

ojeU2Yj.jpg


6mzmVeA.jpg
 
Last edited:

xiaofan

High Supremacy Member
Joined
Sep 16, 2018
Messages
30,969
Reaction score
8,616
Interesting, Starhub now bundles Nokia Beacon 1 router and IPv6 can not be disabled by the customer unless you contact the Nokia customer service to disable it. So looks like Starhub is pretty confident about their IPv6 deployment. :D

Ref:
https://forums.hardwarezone.com.sg/...ic-4/nokia-beacon-1-disable-ipv6-6391405.html

I ended up contacting Nokia, they dibbled it remotely and now it's working great!!! Just contact the Nokia Smarthome helpdesk, they were really helpful!!
 

bert64

Senior Member
Joined
Jan 20, 2020
Messages
1,027
Reaction score
538
M1's seem to have unoptimised route to HE's tunnelbroker server, ping to the server has been hovering around 170ms for the past month or 2.

The server is tserv1.sin1.he.net, interestingly on M1 the ping over ipv4 is around 170ms but over ipv6 it's around 3ms...
Of course since you have native ipv6 on M1 there's little reason to use a tunnel.
 

Mach3.2

Great Supremacy Member
Joined
Apr 8, 2011
Messages
72,405
Reaction score
2,463
The server is tserv1.sin1.he.net, interestingly on M1 the ping over ipv4 is around 170ms but over ipv6 it's around 3ms...
Of course since you have native ipv6 on M1 there's little reason to use a tunnel.
I'm using it to bypass NAT for some of my internal web servers, so it's easier for them to get SSL certs. Then there's also the added bonus of having a block of /48 static IPv6 addresses.

As for the /64 routed from M1, I'm using that for my home network, so there's not really enough addresses to go about without breaking SLAAC. :s22:


Anyway it seem to be back to normal after I got a WAN address refresh to another block of IP.
Code:
Running mtr  -w -c 10 -i 1 216.218.221.42:

Start: 2020-10-20T18:03:06+0800
HOST:                                     Loss%   Snt   Last   Avg  Best  Wrst StDev
  1.|-- 1.128.104.27.unknown.m1.com.sg    0.0%    10    4.1   4.3   2.8   8.4   1.7
  2.|-- 6.246.65.202.unknown.m1.com.sg    0.0%    10    2.7   2.8   2.6   3.1   0.2
  3.|-- 5.246.65.202.unknown.m1.com.sg    0.0%    10    6.8   3.6   2.6   6.8   1.5
  4.|-- 162.246.65.202.unknown.m1.com.sg  0.0%    10    3.5   3.5   3.3   3.6   0.1
  5.|-- hurricane-electric.sgix.sg        0.0%    10   25.9   8.8   3.6  25.9   8.0
  6.|-- tserv1.sin1.he.net                0.0%    10    4.7   6.5   3.5  11.9   3.3
 

bert64

Senior Member
Joined
Jan 20, 2020
Messages
1,027
Reaction score
538
I read it there are no real benefits to the end users but apparently there are benefits to the Telecom operators.

From what I read, China is pushing for ipv6 in the past few years to prepare for 5G and IoT.

Maybe ipv6 will take off after 2025 in Singapore.

There are many benefits of IPv6, both for the providers and the users... HOWEVER, for an incumbent provider that already has enough IPv4 to support their customer base there is a reason to stifle IPv6 adoption as the restricted availability of IPv4 these days makes it harder and more expensive for any new providers to enter the market.

Any new or expanding providers will be forced to implement carrier grade nat (CGN) which is expensive to implement and incurs a performance hit, especially once you have regulatory requirements like having to log every connection in order to identify customers actions.

IPv6 is taking off in many countries, for instance around 70% of users in india have IPv6:
https://stats.labs.apnic.net/ipv6

Less developed and more populous countries are significantly more affected by the flaws of ipv4, and that's why you see a big push for ipv6 in china, india etc.

Big services like Google, Office365, Facebook etc are all available over IPv6 now.

All else being equal, IPv6 is faster - simplified packet header, no need to recalculate the checksum on every router, no need for NAT and its associated overheads etc. How much of a difference depends, on M1 right now my pings to google are around 4ms over ipv4 and 3ms over ipv6, microsoft, facebook etc are similar with ipv6 always being slightly faster. For users forced to use CGN the difference can be much bigger, especially if the CGN gateway is overloaded.

CGN causes other problems, like some other user getting the shared address blacklisted. When forced to use providers with CGN it's not uncommon to be constantly facing captcha requirements from google, cloudflare and other sites.

The Internet was designed on the assumption that every host has a unique address, and a lot of older protocols were designed that way. NAT breaks this assumption, and you end up having inferior workarounds:

  • VOIP - The way VOIP protocols like SIP and H.323 were designed is you have a server to negotiate the call, which then sets up streaming directly between the endpoint devices. With NAT this isnt possible, so the voice traffic gets routed through a third party server too - increasing latency and jitter.
  • Gaming - The same applies, direct connections between gamers instead of via a third party server has lower latency... Microsoft designed the xbox one to support ipv6 partly for this reason.

Latency of a third party server can be an even bigger problem if the other peers are local to you (same country, even same isp) and the server is not (providers of such services will usually choose not to locate local servers in every country, and countries with less users are more likely to lose out here).

NAT also encourages devices to be built to depend on third party services, such as cctv cameras that use cloud services. You can control your own firewall and/or vpn and restrict inbound access to your devices, but you have no control whatsoever over a third party cloud service. These services might be horribly insecure, or might get shut down without notice.

Clearer firewall rules - addresses are the same both sides of the firewall, so your rules are simpler and easier to understand.

Background noise - ipv4 addresses are very limited, so its common for malware to scan the entire address space. This junk will constantly be hitting any box with an unfiltered connection, wasting resources. Such scanning does not occur over ipv6 because the address space is so large it wouldn't be practical to do so.

You may not be forced to use CGN right now, but spare a thought for the millions who have no choice. A lack of widespread ipv6 adoption is hurting the millions of people who have no choice.
 

xiaofan

High Supremacy Member
Joined
Sep 16, 2018
Messages
30,969
Reaction score
8,616
Indeed ping to Google is faster with IPv6, even with Singtel 6rd implementation.

Edit --> not correct. They are the same, as expected.

Guess no real benefits to Singtel 6rd implementation then.

Code:
ping www.google.com.sg
Pinging www.google.com.sg [74.125.200.94] with 32 bytes of data:
Reply from 74.125.200.94: bytes=32 time=6ms TTL=101
Reply from 74.125.200.94: bytes=32 time=7ms TTL=101
Reply from 74.125.200.94: bytes=32 time=7ms TTL=101
Reply from 74.125.200.94: bytes=32 time=7ms TTL=101

Ping statistics for 74.125.200.94:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 6ms, Maximum = 7ms, Average = 6ms

ping -6 www.google.com.sg
Pinging www.google.com.sg [2404:6800:4003:c00::5e] with 32 bytes of data:
Reply from 2404:6800:4003:c00::5e: time=7ms
Reply from 2404:6800:4003:c00::5e: time=6ms
Reply from 2404:6800:4003:c00::5e: time=7ms
Reply from 2404:6800:4003:c00::5e: time=7ms

Ping statistics for 2404:6800:4003:c00::5e:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 6ms, Maximum = 7ms, Average = 6ms
 
Last edited:

xiaofan

High Supremacy Member
Joined
Sep 16, 2018
Messages
30,969
Reaction score
8,616
And just check that SingTel is also using CGNAT for the mobile network even though it is not using CGNAT for Fibre Internet.

The rmnet0 interface of my SingTel mobile has the following info from "ip -a" command under Termux app.

inet 100.106.xxx.xxx/32 scope global rmnet0
valid_lft forever preferred_lft forever
inet6 fe80::xxxx:xxx:xxx:xxx/64 scope link
valid_lft forever preferred_lft forever
 

Mach3.2

Great Supremacy Member
Joined
Apr 8, 2011
Messages
72,405
Reaction score
2,463
It's normal for mobile data users to be not allocated a public routable IPv4 address. My M1, TPG and Grid lines all have non-public routable IPv4 addresses.
 

xiaofan

High Supremacy Member
Joined
Sep 16, 2018
Messages
30,969
Reaction score
8,616
tpg has ipv6 :)

Yes just checked TPG mobile on my Huawei Nova 5T and it actually has a public IPv6 address.

rmnet0:
inet 10.11.xx.xx/32 scope global rmnet0
valid_lft forever preferred_lft forever
inet6 2405:xxxx:xxxx:xxx:xxxxx:xxxx:xxxx:10c/64 scope global
valid_lft forever preferred_lft forever
inet6 fe80::xxxx:xxxx:xxxx:506/64 scope link
valid_lft forever preferred_lft forever
 

xiaofan

High Supremacy Member
Joined
Sep 16, 2018
Messages
30,969
Reaction score
8,616
The server is tserv1.sin1.he.net, interestingly on M1 the ping over ipv4 is around 170ms but over ipv6 it's around 3ms...
Of course since you have native ipv6 on M1 there's little reason to use a tunnel.

For SingTel IPv4 is faster than IPv6.

$ ping6 -c 4 tserv1.sin1.he.net
PING tserv1.sin1.he.net(tserv1.sin1.he.net) 56 data bytes
64 bytes from tserv1.sin1.he.net: icmp_seq=1 ttl=52 time=221 ms
64 bytes from tserv1.sin1.he.net: icmp_seq=2 ttl=52 time=443 ms
64 bytes from tserv1.sin1.he.net: icmp_seq=3 ttl=52 time=363 ms
64 bytes from tserv1.sin1.he.net: icmp_seq=4 ttl=52 time=227 ms

--- tserv1.sin1.he.net ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3002ms
rtt min/avg/max/mdev = 221.090/313.713/443.036/93.891 ms


$ ping -c 4 tserv1.sin1.he.net
PING tserv1.sin1.he.net (216.218.221.42) 56(84) bytes of data.
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=1 ttl=52 time=80.5 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=2 ttl=52 time=80.2 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=3 ttl=52 time=91.2 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=4 ttl=52 time=75.5 ms

--- tserv1.sin1.he.net ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3005ms
rtt min/avg/max/mdev = 75.568/81.908/91.234/5.744 ms
$
 

bert64

Senior Member
Joined
Jan 20, 2020
Messages
1,027
Reaction score
538
Indeed ping to Google is faster with IPv6, even with Singtel 6rd implementation.

Edit --> not correct. They are the same, as expected.

Guess no real benefits to Singtel 6rd implementation then.

Code:
ping www.google.com.sg
Pinging www.google.com.sg [74.125.200.94] with 32 bytes of data:
Reply from 74.125.200.94: bytes=32 time=6ms TTL=101
Reply from 74.125.200.94: bytes=32 time=7ms TTL=101
Reply from 74.125.200.94: bytes=32 time=7ms TTL=101
Reply from 74.125.200.94: bytes=32 time=7ms TTL=101

Ping statistics for 74.125.200.94:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 6ms, Maximum = 7ms, Average = 6ms

ping -6 www.google.com.sg
Pinging www.google.com.sg [2404:6800:4003:c00::5e] with 32 bytes of data:
Reply from 2404:6800:4003:c00::5e: time=7ms
Reply from 2404:6800:4003:c00::5e: time=6ms
Reply from 2404:6800:4003:c00::5e: time=7ms
Reply from 2404:6800:4003:c00::5e: time=7ms

Ping statistics for 2404:6800:4003:c00::5e:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 6ms, Maximum = 7ms, Average = 6ms

The fact that a tunnel is the same as native is pretty bad, what is your pingtime to the 6rd tunnel endpoint? because traffic over the tunnel will always have that added latency... A tunnel should *always* be slower than native.

For tserv1 from M1:

typhoon ~ # ping -4 -c 5 tserv1.sin1.he.net
PING tserv1.sin1.he.net (216.218.221.42) 56(84) bytes of data.
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=1 ttl=56 time=171 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=2 ttl=56 time=173 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=3 ttl=56 time=171 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=4 ttl=56 time=176 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=5 ttl=56 time=174 ms

--- tserv1.sin1.he.net ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 6233ms
rtt min/avg/max/mdev = 171.101/173.120/176.012/1.859 ms
typhoon ~ # ping -6 -c 5 tserv1.sin1.he.net
PING tserv1.sin1.he.net(tserv1.sin1.he.net (2001:470:0:17c::2)) 56 data bytes
64 bytes from tserv1.sin1.he.net (2001:470:0:17c::2): icmp_seq=1 ttl=57 time=3.45 ms
64 bytes from tserv1.sin1.he.net (2001:470:0:17c::2): icmp_seq=2 ttl=57 time=3.21 ms
64 bytes from tserv1.sin1.he.net (2001:470:0:17c::2): icmp_seq=3 ttl=57 time=4.61 ms
64 bytes from tserv1.sin1.he.net (2001:470:0:17c::2): icmp_seq=4 ttl=57 time=3.26 ms
64 bytes from tserv1.sin1.he.net (2001:470:0:17c::2): icmp_seq=5 ttl=57 time=3.65 ms

--- tserv1.sin1.he.net ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4007ms
rtt min/avg/max/mdev = 3.208/3.637/4.613/0.511 ms


Not sure why there is such a difference, the route looks to be the same aside from the last hop so perhaps there's a problem with the tunnel server itself? M1 seem to have direct peering with he.net at sgix...

What sort of throughput do you get too? I find downloading something from https://mirror.0x.sg is a good test as this site is fairly local and dual stack. For the native connection over M1 i get 1-3% better throughput over ipv6.

For some foreign sites i interact with, there is a big difference but this is generally due to one protocol taking a significantly better path.

Assuming native connectivity and the same path ipv6 should always be slightly faster, while a tunnel should always be slower.
 

bert64

Senior Member
Joined
Jan 20, 2020
Messages
1,027
Reaction score
538
It's normal for mobile data users to be not allocated a public routable IPv4 address. My M1, TPG and Grid lines all have non-public routable IPv4 addresses.

It's simply not possible to give routable ipv4 to mobile users, there are simply too many users, and mobile data services came too late in the day.
If you want routable ipv4 on mobile there are specialist providers which will set you up a custom APN and tunnel the session to a server under your control, where you can assign a routable address.

Home internet services traditionally provided routable ipv4, but while providers have enough addresses for their existing customer bases there is no real room for further expansion, so while mature markets typically have routable ipv4 any new providers or developing countries are stuck with CGNAT.

With IPv6 it's normal to allocate a full /64 to each mobile user, your handset will then select random addresses within that /64. If you enable tethering, any of your tethered devices will also receive addresses within the /64 block.

On most providers you also have full two way connectivity, you can connect to your mobile's ipv6 address (or the address of any tethered device) from any other ipv6 capable host on the internet, M1 works this way at least as do most foreign providers while TPG seem to block inbound connections for some reason.
 

bert64

Senior Member
Joined
Jan 20, 2020
Messages
1,027
Reaction score
538
For SingTel IPv4 is faster than IPv6.

$ ping6 -c 4 tserv1.sin1.he.net
PING tserv1.sin1.he.net(tserv1.sin1.he.net) 56 data bytes
64 bytes from tserv1.sin1.he.net: icmp_seq=1 ttl=52 time=221 ms
64 bytes from tserv1.sin1.he.net: icmp_seq=2 ttl=52 time=443 ms
64 bytes from tserv1.sin1.he.net: icmp_seq=3 ttl=52 time=363 ms
64 bytes from tserv1.sin1.he.net: icmp_seq=4 ttl=52 time=227 ms

--- tserv1.sin1.he.net ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3002ms
rtt min/avg/max/mdev = 221.090/313.713/443.036/93.891 ms


$ ping -c 4 tserv1.sin1.he.net
PING tserv1.sin1.he.net (216.218.221.42) 56(84) bytes of data.
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=1 ttl=52 time=80.5 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=2 ttl=52 time=80.2 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=3 ttl=52 time=91.2 ms
64 bytes from tserv1.sin1.he.net (216.218.221.42): icmp_seq=4 ttl=52 time=75.5 ms

--- tserv1.sin1.he.net ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3005ms
rtt min/avg/max/mdev = 75.568/81.908/91.234/5.744 ms
$

What route does it take?
AFAIK singtel peers with he.net in hong kong for ipv4, but in the US for ipv6. I have no idea why they don't peer locally, as HE have a presence in the country and a pretty open peering policy.
 

miloaisdino

Senior Member
Joined
Mar 25, 2016
Messages
843
Reaction score
26
It's simply not possible to give routable ipv4 to mobile users, there are simply too many users, and mobile data services came too late in the day.
If you want routable ipv4 on mobile there are specialist providers which will set you up a custom APN and tunnel the session to a server under your control, where you can assign a routable address.

Home internet services traditionally provided routable ipv4, but while providers have enough addresses for their existing customer bases there is no real room for further expansion, so while mature markets typically have routable ipv4 any new providers or developing countries are stuck with CGNAT.

With IPv6 it's normal to allocate a full /64 to each mobile user, your handset will then select random addresses within that /64. If you enable tethering, any of your tethered devices will also receive addresses within the /64 block.

On most providers you also have full two way connectivity, you can connect to your mobile's ipv6 address (or the address of any tethered device) from any other ipv6 capable host on the internet, M1 works this way at least as do most foreign providers while TPG seem to block inbound connections for some reason.

most mobile providers used to issue public routable ipv4 addresses back in the 3g dongle days.. private ips will be issued to "mobile phone" plans while "wireless broadband" plans were allocated public ipv4s.
 

bert64

Senior Member
Joined
Jan 20, 2020
Messages
1,027
Reaction score
538
most mobile providers used to issue public routable ipv4 addresses back in the 3g dongle days.. private ips will be issued to "mobile phone" plans while "wireless broadband" plans were allocated public ipv4s.

Ahh interesting to know, i wasn't familiar with any mobile providers giving public ipv4 for anything although i never made much use of wireless broadband plans.
 
Important Forum Advisory Note
This forum is moderated by volunteer moderators who will react only to members' feedback on posts. Moderators are not employees or representatives of HWZ. Forum members and moderators are responsible for their own posts.

Please refer to our Community Guidelines and Standards, Terms of Service and Member T&Cs for more information.
Top