Le 10/12/2021 à 10:56, Jiayihao a écrit :
Hi Dino, all,
Based on a user point of view, I try to go through the thread and
summarize the features gathered. Please correct me if anything
missing or I get anything wrong. (Points like 6M mentioned by Fred
are shaped to better reflect the points from users.)
(1) Always-On: be connected to the Internet, Anywhere, by Any links
(either cabled or radio), ALL THE TIME, and All automatically
(without any switch turning). (2) Transparency: be agnostic to the
network protocols (IP, Bluetooth, ZigBee, Thread, Airdrop, Airplay,
or any others), want an easy and straightforward to contact a
people/device without any knowledge of network issues like IP
address, and (3) Multi-homing: seamlessly multi-homing capability for
the host. (4) Mobility: seamless and lossless communications for
moving nodes (vehicle, satellites). (5) Security and Privacy:
security and privacy, omnidirectionally, incessantly (6) Performance:
satisfied (if not impeccable) reliability, availability,
speed(shorter paths/direct communications), enough
bandwidth(10petabit/s for a link), Efficient(less
overlays/encapsulations), highly effective (avoid address waste). (7)
Kernel: make sure the Internet does no harm. (8) Others: no worry
about MTU
Thanks for listing the "Internet - no harm" point.
But I did not see why is it called 'kernel'?
I was thinking to make sure Internet does no harm in the linux kernel
sense, yes, but among more other aspects.
"Internet - no harm" point in the linux kernel would mean probably
something to reduce the size of the Internet (IP) stack in the kernel,
reduce its energy consumption, reduce the 'software bloat' of it. Is
this the kernel you refer to?
"Internet - no harm" point on a broader scale would mean to try to make
sure aspects such as datacenter energy consumption (there were no such
big datacenters prior to Internet; it can safely be assumed that these
datacenters are created and needed by the Internet), human factors
related to over-use of the Internet like professional emails during
weekends and burnout, societal impacts digital divide like
all-Internet-for-young and nothing-for-the-elderly or like the divide
between countries with differing revenues per capita, social system
divides like Internet ability in certain countries to be 'cut', various
legislation's crime facilitated by the Internet, and more - are fed back
into the design of the new Internet and make sure it does no harm.
It is a little bit like in Health and medicine: first, make sure do no
harm. When a surgeon opens someone's body, the first thing s/he must
make sure is to do no harm (do not break some artery or vein, and just
have a look to see what's wrong in the ill), even though the cut in the
skin is already a little bit of harm.
Alex
Thanks, Yihao Jia
-----Original Message----- From: Dino Farinacci
<farina...@gmail.com> Sent: 2021年12月8日 21:36 To: Jiayihao
<jiayi...@huawei.com> Cc: int-area@ietf.org Subject: Re: Side meeting
follow-up: What exact features do we want from the Internet?
[jiayihao] Here I mean our data is controlled by our service
provider not user themselves. For example, if I bought a movie X
from an online streaming provider A, I still do not own this movie
X and have to pay again if I want to watch movie X from online
streaming provider B. However, I am not that sure if it can be
categorized to a network requirement.
This is certainly an application specific issue. Not network layer.
But if you are watching a movie and move from your house to your car
and switch from wifi to LTE/5G, the movie should continue playing.
That is where the network layer provides seamless connectivity and
the app doesn't know that anything has changed (and hence why the
apps use sockets that bind to non-topological addresses (i.e.
EIDs)).
(4) Similar to your first point: I want to be always attached to
the Internet by any personal devices. (then I can better enjoy the
features above)
Yes.
Here I have another question: application developer/programmer
should have a different angle for new features. Of course
application developer/programmer should to be agnostic (or learn
as less as possible) to network stuffs during developing, but
definitely they have more insights and more details shall be
learned by them.
App developers control user behavior by making their
services/functionality available with good UI. Make it easy for the
user to get do things with the fewest number of steps. A counter
example is the Facebook iOS app.
[jiayihao] Agree that App developers should not care about network
stuff like the session continuity during device movement. However,
App developers do learn things like URL, domain names, DNS, Quic,
CDN, proxies.....,and make them transparent behind the good UI. So
there do have some network stuffs that are agnostic to users but
are not agnostic to App developers.
Right, and they should bind/map to EIDs. Meaning, when application
handles or directories map to IP addresses, they are not topological
addresses.
[jiayihao] My observation is: If we consider network
features/requirements from App developers, it will finally result
in discussion about the network architecture.
No, my experience is that apps bind to names. The app developers have
enough fish to fry, they don't want to deal with network specific
issues. They just want to out sockets to send and receive. Just have
a look at all the new Dapps being written.
Dino
_______________________________________________ Int-area mailing
list Int-area@ietf.org
https://www.ietf.org/mailman/listinfo/int-area
_______________________________________________
Int-area mailing list
Int-area@ietf.org
https://www.ietf.org/mailman/listinfo/int-area
