Slow WiFi with Android Marshmallow & IPv6?

Mon Apr 25 15:21:11 CEST 2016

The DNS version of HE doesn’t need to penalize IPv4, so it doesn’t need to add latency, though you’re right, it will increase the number of DNS queries.  That said, how much more network traffic will DNS HE add in comparison to the HTTP HE that exists in today’s web browsers?  My gut says that the number of DNS queries is less than the number of HTTP connections, though that could change with HTTP 2.0.  

In the end it’s all about helping end-users have a great experience despite their infrastructure … let’s not punish the end-users because of all the spaghetti. 

Frank

From: Lorenzo Colitti [mailto:lorenzo at google.com] 
Sent: Sunday, April 24, 2016 10:45 PM
To: Frank Bulk <frnkblk at iname.com>
Cc: IPv6 Ops list <ipv6-ops at lists.cluenet.de>; Erik Kline <ek at google.com>; Eric Vyncke (evyncke) <evyncke at cisco.com>; Jeroen Massar <jeroen at massar.ch>; Brzozowski, John Jason <jjmb at jjmb.com>
Subject: Re: Slow WiFi with Android Marshmallow & IPv6?

All happy eyeballs algorithms have user impact in terms of increased latency, increased load, or both. We must always ask ourselves whether the increased latency/load is worth the benefit in response time on broken networks.

On Mon, Apr 25, 2016 at 12:24 PM, Frank Bulk <frnkblk at iname.com <mailto:frnkblk at iname.com> > wrote:

I thought it was a pretty standard DNS client implementation to move past unresponsive DNS servers.

We hand out several internal DNS servers to our corporate endpoints and I really hope they will quickly try the next DNS server if one server happens to be down for whatever reason.  

Windows: See “Querying the DNS Server, Part 3” https://technet.microsoft.com/en-us/library/dd197552(v=ws.10).aspx

OS X: https://support.apple.com/en-us/HT203244

I would hope that same approach would be used in other operating systems, regardless of the transport mechanism (IPv4 or IPv6).  In fact, I would like to see a Happy Eyeball implementation in DNS, too.  I don’t want to see service provider or corporate helpdesks telling customers to disable IPv6 just because there’s some non-optimized IPv6 deployments out there – both the helpdesk agents and customers will have hard time unlearning this behavior.  

Frank

From: ipv6-ops-bounces+frnkblk=iname.com at lists.cluenet.de <mailto:iname.com at lists.cluenet.de>  [mailto:ipv6-ops-bounces+frnkblk <mailto:ipv6-ops-bounces%2Bfrnkblk> =iname.com at lists.cluenet.de <mailto:iname.com at lists.cluenet.de> ] On Behalf Of Lorenzo Colitti
Sent: Sunday, April 24, 2016 7:36 PM
To: Brzozowski, John Jason <jjmb at jjmb.com <mailto:jjmb at jjmb.com> >
Cc: Erik Kline <ek at google.com <mailto:ek at google.com> >; Eric Vyncke (evyncke) <evyncke at cisco.com <mailto:evyncke at cisco.com> >; Jeroen Massar <jeroen at massar.ch <mailto:jeroen at massar.ch> >; IPv6 Ops list <ipv6-ops at lists.cluenet.de <mailto:ipv6-ops at lists.cluenet.de> >

Subject: Re: Slow WiFi with Android Marshmallow & IPv6?

Android does not behave well when configured with IPv6 DNS servers that do not work. This is because it prefers IPv6, does not (yet) ignore unresponsive DNS servers, and has quite high resolver timeouts.

One infamous example is a German ISP whose CPE announces an IPv6 address in RDNSS but never responds to queries to that address. Perhaps because no other operating system has a problem in this situation, the ISP has not fixed this, and users have been blaming Android. This is by no means the only case, though. Some of you know who you are :)

Future Android releases will likely ignore broken DNS servers. This is unfortunate; we'd hoped that ISPs that provision devices with IPv6 connectivity would be able to ensure that the DNS servers are responsive and that as IPv6 matured this problem would go away. Unfortunately it has not.

As Erik said, the way to debug this problem would be to have someone running 6.0.1 on voo to run "adb shell dumpsys connectivity --diag" and open a bug as described at https://source.android.com/source/report-bugs.html .

On Sun, Apr 24, 2016 at 11:21 PM, Brzozowski, John Jason <jjmb at jjmb.com <mailto:jjmb at jjmb.com> > wrote:

Android is properly trying to query DNS over IPv6 unlike other devices/OSes.  Most other mobile platforms still prefer the querying of DNS over IPv4 for A/AAAA RR query types.

On Sun, Apr 24, 2016 at 10:17 AM, Eric Vyncke (evyncke) <evyncke at cisco.com <mailto:evyncke at cisco.com> > wrote:

Jeroen, Erik and John,

Thanks for the hint. I will advise the ISP to investigate any DNS issue (such as not returning an error message when requesting a non-existing AAAA) but I wonder why it is linked to that specific Android Marshmallow version.

-éric

From: <ipv6-ops-bounces+evyncke=cisco.com at lists.cluenet.de <mailto:ipv6-ops-bounces+evyncke=cisco.com at lists.cluenet.de> > on behalf of "Brzozowski, John Jason" <jjmb at jjmb.com <mailto:jjmb at jjmb.com> >
Date: Sunday 24 April 2016 at 16:01
To: Erik Kline <ek at google.com <mailto:ek at google.com> >
Cc: Jeroen Massar <jeroen at massar.ch <mailto:jeroen at massar.ch> >, IPv6 Ops list <ipv6-ops at lists.cluenet.de <mailto:ipv6-ops at lists.cluenet.de> >
Subject: Re: Slow WiFi with Android Marshmallow & IPv6?

My customers saw this issue at one point.  We had issues with DNS over IPv6.  Bad DNS and/or network configurations.  Once these were fixed, the problems cleared up.

On Sunday, April 24, 2016, Erik Kline <ek at google.com <mailto:ek at google.com> > wrote:

On 24 April 2016 at 19:53, Jeroen Massar <jeroen at massar.ch <mailto:jeroen at massar.ch> > wrote:
> On 2016-04-24 11:51, Eric Vyncke (evyncke) wrote:
>> One of the first Belgian ISP to deploy IPv6 (VOO) is now recommending to
>> its Android Marshmallow (6.0.1) users to deactivate IPv6 on their
>> residential WiFi CPE... :-(
>>
>> It appears that the issue is about IPv6 web sites/apps being really
>> slower when using IPv6.
>
> Is it a DNS issue maybe?
>
> https://www.sixxs.net/faq/dns/?faq=ipv6slowconnect
>
> As that has been the general cause of "Disable IPv6!!!!!" around the
> world for many years already.
>
> Of course, without more details, little one really can say. Bug number
> maybe?
>
> Greets,
>  Jeroen
>

Yeah, a link to something that eventually leads to a bug report would be good.

Also if anybody has adb installed they can just try "adb shell dumpsys
connectivity --diag" and see what the over-simplified diagnostic
output shows.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cluenet.de/pipermail/ipv6-ops/attachments/20160425/b4dd9128/attachment.htm>