Anyone familiar with RINA (an alternative network ...
# thinking-togethers
shalabh
03/10/2019, 8:57 AMAnyone familiar with RINA (an alternative network architecture)? https://en.wikipedia.org/wiki/Recursive_Internetwork_Architecture
shalabh
03/10/2019, 8:57 AMThere's also http://www.martingeddes.com/think-tank/network-architecture-research-tcp-ip-vs-rina/.
Excerpt:
The core insight behind RINA is the observation of a simple recurring pattern in all of distributed computing. “Communicate this for me from here to over there” is a ‘what’, which is then followed by a bunch of common functions that are the ‘how’. Those ‘how’ functions relate to dividing the data stream up into datagrams and reassembling them. That can be done in any way the lower layer sees fit, subject to the contract it has with the upper layer.
p
Peter Abrahamsen
03/12/2019, 12:35 AMI hadn’t heard of it, and reading that post told me more about Geddes’ ego than RINA
Peter Abrahamsen
03/12/2019, 12:41 AMIt sounds right that IPv6 is solving the wrong problems, though. Networking is another area where there’s so much complexity, so much history, so many deployed systems and so many actors, that we can only solve problems in ways that add inertia.
Peter Abrahamsen
03/12/2019, 12:44 AMI would fully expect a full reimagining of networking to be dramatically simpler than TCP/IP, knowing nothing of the particular design.
Peter Abrahamsen
03/12/2019, 12:46 AMThe Wikipedia article also reads as a diatribe. Ugh.
Peter Abrahamsen
03/12/2019, 12:58 AMIt feels really important to figure out how to be better at building systems that can work at scale and still evolve. Despite good intentions, there are always elements of a design, or of the real world system that develops around it, that eventually need to change, and can’t. Rallying against this kind of problem is maybe one of the core things unifying Future of Programming projects.
❤️ 1
b
Breck Yunits
09/04/2019, 11:09 PMI don't think I've seen RINA before. Thanks. I'm not a network architecture expert and so don't have strong knowledge of the problems with the network topology. I am interested in this space, however, as I have worked a lot on an idea to create a network stack that uses 1 universal syntax for all protocols (replacing TCP/IP, HTTP, etc) now called the World Wide Tree (in 2013 I originally called it SpaceNet https://medium.com/space-net/spacenet-51aca95d49a2). Perhaps there are some synergies there
Breck Yunits
09/04/2019, 11:10 PM(I should say 2 universal syntaxes--binary is still at the lowest level, of course)
s
shalabh
09/05/2019, 6:06 PMYeah I feel that, as great as the internet is, the protocols used are kind of a hodge podge of overlapping ideas and re-implementations. Well, nobody reimplements TCP/IP which is the substrate. But it provides just unreliable datagram or realiable byte streams (why realiable datagrams, e.g. SCTP, isn't used everywhere?) Almost all protocols I've seen reinvent enveloping and reliable datagrams on top of the streams in their own way, because you typically need messages, not streams. Another thing that's reinvented is how to represent structure in bytes, similar to file formats. I think the deeper, and harder problem is location and naming of resources. There's DNS but it's not sufficient, so HTTP adds another layer (the url) to name more resources and so on.
👍 1
shalabh
09/05/2019, 6:14 PMI don't remember the details but Joe Armstrong once talked about a protocol that had general s-expr style structure built in and you could use it to send any kind of structured messages across so 'in theory' HTTP, SMTP etc could be designed on top.
👍 1
b
Breck Yunits
09/06/2019, 2:28 PMperhaps this https://joearms.github.io/published/2016-01-28-A-Badass-Way-To-Connect-Programs-Together.html (there's a great Alan Kay quote on there too)
3 Views