Fredy Kuenzler: Buffering sucks!

preroll music

Swiss German would be an option for me.

English, because you know the
Swiss don't speak proper German.

My six year old digital native
is telling people rather proud

that his Dad invented the
fastest internet in Switzerland.

It’s called Fiber7.

applause

Thank you.

While we went to Greece for vacation, I
was in a target conflict, because I had to

explain him why he couldn’t watch YouTube.
I mean Greece, you know it’s maybe a bit

difficult, but as a matter of fact, here
in Hamburg it’s not any better. I’m next

door in the hotel InterCity and they offer
“free WiFi” with 256 kbit/s. If you want

5 Mbit Internet, you pay 8 Euros extra,
per day. So this is where we are in 2015.

A few words about me: I’m married, one son
as I said. He was born 2009. He was able

to unlock the iPhone with the age of 17
months. No one showed him how. My early

connection with digital techniques was
about 1978 when I was playing with these

chips 7400. Who knows them? Raise your
hand. Few, thanks. Later on I did an

apprenticeship as a Fernmelde- und
Elektronikapparatemonteur. And I started

to do IT business about 1991. And 1996
– almost 20 years ago – we started with

Linux stuff. My first Linux was Suse 4.2.
In the year 2000 we started with Init7

and later on I became president of the
SwissIX association. This is an

association which runs a Internet
Exchange. I had also my time in a startup

called Zattoo. It’s a network architecture
OTT IP Television. Besides, I need a

hobby, so I’m also a politician for the
Social Democrats in my city parliament,

already 8 years. Then I started with the
other hobby, Fiber7 as you know. Oh

besides, I was also working in an internet
expert group of the Social Democrats

Switzerland. The internet paper was
adopted earlier this month by the national

Delegiertenversammlung. I don’t know what
this is in English. So, Buffering sucks!

Ladies and Gentlemen, this talk is not
about Deutsche Telekom. It’s not about

peering. It’s not about interconnection.
It’s about these thousands and millions

of youngsters out there which want to
watch YouTube in HD resolution without

buffering. So let’s quickly look at the
reason why YouTube and all the other video

buffers. It’s usually lack of bandwidth.
If you have a 2 Meg DSL or if you have

an InterCity free WiFi with 250 kilobits;
so HD video is not possible. Sometimes

they have old PCs, so CPU power is an
issue – these days no longer relevant.

WiFi quality sucks sometimes. This is
rather an individual issue. And sometimes

we have an over-subscription of the shared
node – mainly in cable networks. Streaming

source can be too far away. If you stream
from the U.S., it doesn’t really go well.

That’s why we have so many CDN, Content
Delivery Network systems, close to the

end users. Then adaptive streaming can be
an advantage, but also disadvantage. You

cannot turn it off. When you watch HD and
the connection sucks you just cannot keep

it on HD. It just drops to SD or lower
resolution. It works, yes. But Claire

Underwood in low res is not so cool.
Routing algorithm issues – sometimes it’s

a mismatch of client and server. If your
client is assigned to the wrong CDN

server, then it’s also slow. Anycast
routing is a trick sometimes. Last but not

least and the most important thing: It’s
over-subscribed interconnections. We go

back quickly to the old days. The caller
pays. When you call your mother-in-law

and you talk with her – well, she talks to
you for 45 minutes and you say hello and

goodbye – you still pay the call.
laughter

So with YouTube it’s not any different.
You click YouTube and then YouTube talks

to you for hours maybe and then you say
goodbye, basically. So is the broadband

customer calling the YouTube server or is
it vice versa? Is the YouTube server

calling the broadband customer? Probably
it’s the broadband customer who calls.

But still the data is flowing from the
server to the client. But the client is

causing the traffic, because he is
requesting the traffic. And if we look at

the structure of the internet, we have
basically the end user to the right. We

have – here is the provider network and
the end user is only connected to the

provider’s network. On the left side we
have all the content in the internet.

We have the media and video and streaming
and Torrent and you name it. But there is

always only one way going to the end user.
It’s the yellow marked interconnection

points and there is no way around them.

This basically means, the provider
can monopolize the end customer.

At least as long [as] he is
connected or subscribed.

There is no alternative way.
So this gives the provider

a position of power. On the other hand

these interconnection points used
to be – for a long period of time –

so called Zero Settlement interconnections
and they are basically the foundation of

the internet. Without Zero Settlement
peering, without interconnection the

internet wouldn’t exist as we know it.
The broadband provider, mainly the

incumbent, the ex-monopolist, or large
cable operators, they tend to become

more and more restrictive to provide
sufficient interconnection capacity.

Not upgrading interconnection to the
requirements is very common these days

and it’s a passive aggressive
behavior. So many providers

– to name a few: Deutsche
Telekom – they just do nothing.

They just wait. And the end customers
are suffering. Buffering is very common,

especially during prime-time.
This is basically what the topic of…

…the main topic of this conference is:
It’s a gated community. The provider

creates a gated community
for his own end customers.

So as I said before: The data
is flowing from the server,

from the video server to the end customer.
It’s about 50 times more traffic

flowing to the client and the
usual traffic ratio we have

for a broadband provider is 1:5
or 1:10. So they’re pulling about

10 times more traffic
towards the end customer.

Then we have this interconnection
policy. So they don’t do anything.

As I said before, they just
over-subscribe the existing

interconnection. And if you want to
upgrade you have to have a traffic ratio

of about 1:1.5 to 1.3. But no video
stream service can deliver traffic and

also maintain the traffic ratio. No
content provider can. So all they can do

is: They can pay money to get upgraded.
And if they don’t pay, data is stuck in

congestion and the clients are suffering,
seeing the buffering sign. Large broadband

providers, such as the incumbents and
cable providers, they want to get paid

twice. They are able to force the money
due to the temporary monopoly – as I

explained. And they can ask money from the
end customer and on the other hand also

from the content. This is called double-
sided market. And if they don’t pay, the

content is not paying, this is what we
see. And sometimes – as a side note – the

end customer pays, but still sees this.
But IP interconnection would be cheap.

The business cost per broadband customer
is just a few cents per month. And if the

provider would invest this, people would
be happy. On top content providers are

easy to deal for peering or provide cache
servers etc. So please talk to our

community fellows of Akamai, Apple,
Amazon, Facebook, Google, Limelight,

Netflix. T is not Telekom, it’s Twitch.
And Zattoo and a lot of others. So traffic

congestion is costly. I took a random
Google search and was looking for how much

traffic is actually costing. And “Die
Welt” showed the result: “Staus kosten

in jedem Haushalt 509€/Jahr”.

So my assumption was: If traffic
jam is costing money, then

probably data traffic jam is also
costing some money. But I figured that no

one was really exploring that field, yet.
So I thought I’m going to do a little

“Milchbüechlirächnig”
laughter

applause
When I was a child, the milk man came

every morning and we just put our order
into the Milchbüechli and he put the milk

into the box outside of the house. By the
end of the month, we went to the shop

and paid our Milchbüechlirächnig. So this
is my quick calculation: We have about 30

million broadband connections in Germany.
I assume that everybody is waiting for one

minute accumulated while watching Netflix,
YouTube, whatever. Probably this is far

too less. Who thinks one minute is fine,
or – who thinks one minute is not enough?

Oh, ok, so let’s stick with one minute for
the calculation. And I also assumed that

5€ / hour waiting is a good salary. If you
think, 5€ is not enough, you can adapt the

calculation. This is called
“Reservationslohn”. I have no clue what it

means, but this was on Wikipedia, for
time when you take a job or refuse a job,

how much would be the value for the
spare time. So this is my calculation:

If you wait one minute per day, this is
6 hours per year. If you multiply this

with the 5€, every broadband
customer would lose 30€ per year.

This sums up – with 30 million
broadband subscribers –

to 900 million Euros per year. This is the
economic damage in Germany per year.

applause

As we can assume that a large part of the
buffering is caused by the insufficient

interconnection, especially during prime-
time when everybody wants to watch

Netflix. This is also a result of
the restrictive peering policy

of the incumbent and large cable operators
and the ability for them to force

some extra money out of these double
sided market power as I explained.

They probably would gain a few millions.
I don’t have exact figures but I assume

it’s probably some 10..20..30
millions per year, they could force

through this market power. On the
other hand we have the damage

of 900 Million Euro per year and I mean
this is like a – how do you say that? –

imbalance. So my conclusion in democratic
countries like [in] Western Europe:

The economic gain of a multibillion
company at the expense of

the general public is commonly not
tolerated. The next question is basically

following the previous talk of Thomas:
When will the regulators wake up and force

every market participant to cooperative
peering and interconnection because the

end user is suffering, the public is
suffering. Zero Settlement peering – as I

explained – is rather common. Of course
the “Deutsche Telekom Lobby” would tell

otherwise, this is clear. The unbalanced
traffic should no longer be used to refuse

peering; and also disputes about the
interconnection should be resolved rather

quick. My case against Swisscom is taking
years already and still no end and no

light at the end of the tunnel. Then, last
but not least we should have broadband

providers, must be committed to the
interests of their own end user customer

base. As I said, Telekom managed to get
paid twice because of their market power;

and other Telecoms, such as Telecom
Hungaria or Swisscom, they use Deutsche

Telekom and their market power as a
leverage to force their also restrictive

peering policy; and the regulators so far
don’t do much. I quote here Marc Furrer,

this is the chief of ComCom Switzerland:
“Nur ein fauler Regulator ist ein guter

Regulator”.
laughing

Thank you! Questions?
applause

Herald: Okay, thank you Fredy; and let’s
have Thomas back up on stage and we’re

gonna take questions, please. There is
actually more than the mics I said before,

there is 2 right up on the top and there
is 3 in each aisle. So if you please

line up if you have any questions and ask;
and please speak into the mic, we need

your questions on tape, and those who
are leaving now: Do it silently please.

Okay, first question, over there!
Q: I have a question for Thomas:

From your talk it sounds like you did a
lot of work. Can you tell us a little bit

about the budgeting, that goes
into having a team like that?

T: Yeah, so, SaveTheInternet
is a coalition of 12 NGOs

which have all their independent budget.
There is no fixed budget for the work

that we have been doing as a whole.
All of them have transparency reports.

So I can not really speak for the
budget of EDRI or accessnow.

The organization where I am
based in Austria got a grant from the

media democracy foundation from 10.000€
and money from Netflix, 10.000€ also; and

we used both for development and paying
for the Faxes. Because in the second run

of the Fax tool the provider that it
was referring to was no longer paying.

Otherwise the funding in general about
Digital Rights in Europe is awfully low.

So if you compare it to the U.S. where you
had double-digit millions going into the

lobbying it is ridiculous what resources
we have here in Europe; and we are

thinking about making a donation tool for
the new SafeTheInternet; but again that’s

complicated because you have 12 NGOs
with very different activity scales. Like

some of them do a lot, others not so much.
So how would you divide the money?

These are unresolved questions, that we
are working on right now. If you wanna

support us with independent
funding, then just donate to the

individual organizations.
EDRI, Initiative für Netzfreiheit,

are probably the ones I would mention
most, because they have done

most of the work; accessnow as well,
but they generally have a lot of funding

from the U.S., so I don’t think
they need it that much.

Q: But to summarize, I saw a picture of
your team. I saw all the work you did.

You did that for 20.000€?
T: No. I never got a Cent. I was paid by

EDRI for 4 months when
I was working in Brussels

within BEREC for the first reading;
but otherwise this was mostly free

time. I got my expenses covered for travel
but other than that I am doing this in my

spare time. Also now unemployed…
applause

I work for Data Protection NGOs, so they
are allowing me to do a lot of my stuff

also for Net Neutrality.
Herald: We’re all elephants. We do it

for peanuts. Okay, No.1 go ahead!
Mic 1: Yeah, hello! Hi Thomas, thanks a

lot for your work, that’s great. I have a
question about the involvement of the

Business Angels and the companies:
What is the reason, what do you think why

they came so late into this discussion in
Germany. What probably can we do to change

this in the future because I think that’s
a… they are great allies in this fight.

Thomas: That’s… you’re asking exactly the
right question. Sadly in Europe you have

no organized voice for Startups
or for SMEs when it comes

to Digital Rights issues; and you would
have to work with them to get them

involved in the debate. They were really
late to the party and then, again,

mostly activated through
U.S. networks. So the

connection between the civil rights scene
here and the business scene, particularly

the one which is organized in Brussels
with European umbrellas is very weak. So

everything you can do there to
strengthen this connection would be great.

But I don’t have those business contacts.
I got a few people involved in the first

reading stuff but we definitely need more
people that act as multipliers to get more

companies involved, particularly now when
we enter into a new phase with the BEREC

guidelines. We no longer need the loud
arguments of many people, we need more the

arguments from the business side, from the
universities, from those people who run

networks. These arguments are better
suited to make a difference with the

regulators.
Fredy: And to add: Don’t underestimate

the influence of the lobbies, of the big
names, the Telecoms and the liberty

globals, they have a lot of money and they
try to influence the politicians as good

as they can. They do a good
job from their perspective.

Thomas: You can be sure that the Telecoms
will have people for all 28 regulators,

now continuously lobbying for an upcoming
9 months. The question is: Who is in our

team?
Herald: Okay. Thank you. Is there a

question from the internet?
Signal Angel: Yes, there is a question,

it is: Whether peering providers should
differentiate between virtual private

network traffic and public traffic and
where is the line between internal network

and the public internet?
Fredy: What should I say… this is

difficult question, I mean… Basically, if
you all commit your backbone then

there is always plenty of traffic… or
plenty of capacity. So there is… there

shouldn’t be any differentiation. Networks
should provide enough capacity and then

we’re good. A common argument from the
big names: “Oh we are investing millions

and millions and millions in broadband
expansion” but unfortunately they stop

investing right at the end of their own
backbone so they don’t invest any money

beyond their little percentage of the
total investment for their interconnections.

Herald: Okay, there is another question
at No.1?

Mic 1: I have a question about buffering:
So the most of the content in the web is

delivered over TCP/IP and… will changing
the media to something like UDP which has

lower overhead over TCP/IP;
will that change the situation?

Fredy: Not really.
Mic 1: No?

Fredy: No. It won’t help. I mean packet
loss is packet loss regardless whether it

is TCP or UDP.
Herald: Okay, that was a short answer. Next

question please. Please talk into the mic.
Mic: So when I came here, this year,

I had the impression that at digital
subscriber line connections not only

bandwidth is bad but also the ping gets up
way high. Of course – I mean – at home I

have Fiber7 nowadays so I just thought
I got spoiled by fiber connections but I

noticed that ping times went up from,
well, couple of years ago 60-80ms from

sites in your neighborhood more or less
to nowadays 80-160ms. Where is the

problem there?
Fredy: Well the latency is directly

related if the provider is not delivering
enough bandwidth, then ping goes up

that’s a normal behavior of TCP.
Mic: So the problem is also at the

interconnection sites?
Fredy. Probably yes, most likely, you

can find out if you do traceroute, then
you see where… well, there is a long

presentation how to interpret traceroute
properly. If you look for “Nanog traceroute”

you should find this lecture. But that
would probably give some indication.

Mic: Alright, thank you.
Herald: Thank you. Next question from

the internet, just in between and
then we’ll go back, go ahead.

Signal Angel: “Is Netflix a gated
community by itself?” and “Are you sure

that their interest will align with the
movement of net neutrality in the long

run?”
Fredy: We should differentiate between

Netflix content and Netflix interconnections.
So for the content I probably

would say: Yes, but I am not the expert.
This would be then layer 7 in the OSI

model. I am talking here on layer 3, this
is content agnostic. Netflix, they are one

of the good guys because they really help
to deliver the packets. I know them

personally a few fellows from the peering
community. They are the good guys,

definitely.
Thomas: Just also to answer this question

for the European debate, Netflix was one
of the good guys in the U.S. and they also

supported of course the European movement.
But again, they are so big that I wouldn’t

really trust them as an ally because they
could also pay, they could also survive in

a double sided market and for them in the
growing emerging markets like Europe where

they just have started, it’s probably
risky to allow for this new type of anti

net neutrality business models; but in the
consumer side where net neutrality is seen

as an end user issue I think so far their
interests mostly align. On interconnection

they have their own interests of course.
Fredy: So I can say: Netflix is

definitely paying Deutsche Telekom
otherwise no single Deutsche Telekom user

would be able to watch any
movie on Netflix! So! For sure!

Herald: Okay, we are short for time
so please, last 2 questions. No.2

first. Keep it short please.
Talk into the mic.

Mic 2: Regarding the first talk: What is
the… do you have an explanation for the

behavior of the European Commission
in behave of the net neutrality debate?

I especially think of the behavior of
Günther Oettinger who repeatedly said his

ridiculous lie of a “net neutrality kills”
and he repeated it again and again

even if there was no reason behind
it. And do you have an explanation for

this behavior of the Commission
and Junker and this.

Thomas: For that argument, we had this
great YouTube video “net neutrality kills”.

If you search it you will find it or
“Netzneutralität tötet” in German. That

deconstructs this argument of Oettinger.
But in general, and you can go back to the

previous commissioner Neelie Kroes that I
showed. Our sole suspicion is that the deal

was that the telecom industry has to give
up a little bit of their profits when it

comes to Roaming, but on the other
side they gain a lot of future profits on

the abolishment of net neutrality and so it
was like “Okay, we need a Populist argument",

Neelie Kroes also needs a quick win at the
end of her career and this was again like

you take a little bit there and put it
there for the Telecoms industry. And

Oettinger is a big industrial favor guy,
he is always for big business.

Herald: Okay, short for time,
last question, No.1.

Mic 1: Hi, so what strategy should an ISP
use when their capacity on their backbones

is fully loaded? Like first-in-first-out
or what is your idea about that, because

the capacity is limited, so when there is
so much traffic that everything is stuck.

Fredy: Upgrade!
Thomas: Yes, invest in the network!

Fredy: I mean, sorry, a 10G port is now
some 3000€ including optic and cross

connect. It’s not that much. Upgrade!
Herald: Okay, thank you!

applause

postroll music

Subtitles created by c3subtitles.de
in the year 2016. Join and help us!