[ih] Preparing for the splinternet

[Toerless Eckert]

Karl,

How is what you are describing not what we already have for decades ?

Even in the 90th and 200x, we had IP networks of global companies with
as much as 100,000 routers (in 2005 from one corporation if i remember correctly) or
government agencies like the US defense networks, all of which had tightly filtered
connections to the Internet. Who would today dare to put arbitrary home equipment you
buy "onto" the Internet without a strong firewall ? Aka: the end-to-end Internet
was and is mostly a great marketing message up to the point where whoever asked starts to
worry about attacks/security/privacy and so on.

The only somewhat novel aspect is more and more state actors wanting to filter
the public Internet. Including the EU. And there is and probably never will be
a more differentiated answer from IOSOC than "don't" and then being quiet when
being challenged on that ultra-simplified position.

But to me all that splinternet filtering is a secondary issue. I am more worried
about the opposite: The drive for lower cost has eliminated more and
more of such large "private" networks and replaced their backbones with "SD-WAN", aka:
tunneling across the Internet. I have seen no good analysis/prediction of resilience
under failure or attack of/against the Internet and the impact not only for 
"native?" Internet services, but everything that tunnels across it. Sure, the Internet
has likely more redundancy now than 15 years ago, but does that compensate for loss
of independent underlying physical connectivity we had in before ? arguably, even those
private networks might have shared fibers.

Aka: How much insight do we have about the history/evolution and current state of resilience
at the actual application/service level ?  To me even the fact that we might not have enough
 of an idea would be a good reason for DK to send out the mail he did. But even more, it
would be a good reason to better understand/analyze and improve the situation.

Cheers
    Toerless

On Sat, Mar 12, 2022 at 01:56:51AM -0800, Karl Auerbach via Internet-history wrote:
> On 3/10/22 5:02 PM, the keyboard of geoff goodfellow via Internet-history
> wrote:
> 
> > EXCERPT:
> > 
> > According to Wikipedia
> > <https://en.wikipedia.org/wiki/Splinternet#:~:text=The%20splinternet%20>, a
> > researcher at the Cato Institute first used the word "splinternet" in 2001
> > to describe the idea of "parallel Internets that would be run as distinct,
> > private and autonomous universes."
> 
> Well, "splinternet" it isn't quite "Internet history", it's more of a
> prophesy of things that could come.  And I sense that none of us want
> disjoint "splinters", and I don't think users want that either.  But if we
> use the analogy of wood then today's Internet is nice, clear lumber. And
> what we have called "splinternet" might be gluelams or fiberboard, i.e. many
> pieces that are joined to form something that is at least as strong and
> useful as a timber cut from a single tree.
> 
> If we expand our view of Internet History to encompass predecessors we see
> that that "splinters" have existed yet the system as a whole provided
> acceptable service to users.
> 
> Perhaps the earliest system that used store-and-forward handling of
> electronic messages was the telegraph system that arose in the 1830s. 
> Although it was never a single technically uniform global system, it did
> have "splinters" that worked acceptably well and were sufficiently joined so
> that from the users' point of view, it was one system.
> 
> (We can say the same about the voice telephone system, but I view that more
> as a circuit switching paradigm rather than store-and-forward message
> handling.)
> 
> I, personally, am of the belief that just as the Internet began as a single
> network and then became a network of networks, i.e. an Internet, the time
> may be near when we add yet another tier; that the Internet evolves into a
> network of internets.
> 
> How this may come to pass is uncertain.  However, I believe that the weak
> fracture plane is that users no longer care about elegant end-to-end
> principles but, rather, live in a world of Apps and those users care nothing
> whether the underlying plumbing is elegant or a jumble - the users only care
> that their favorite Apps work.
> 
> Early Internet protocols needed end-to-end connections.  But as the years
> passed more and more protocols were designed with the idea that they could
> operate via relays and proxies.  SMTP was an early one, HTTP a later one. 
> It is that acceptance of proxies and relays that reduces the strength of the
> end-to-end principle to act as a glue that holds the Internet into a single
> system.
> 
> (This is not a negative reflection on those protocols; it is merely the
> recognition that a useful and common "feature" may also become the means
> through which the net could separate into realms that touch one another only
> via relays and proxies.)
> 
> I wrote about this some years ago in note I titled "Internet: Quo Vadis
> (Where are you going?)" at
> https://www.cavebear.com/cavebear-blog/internet_quo_vadis/
> 
>         --karl--
> 
> 
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history

-- 
---
tte at cs.fau.de