---
tags: breakout
---

# Breakout group suggestions

1. proxies/in network processing and content owner versus delivery
* v....tensions in distributed/federated (dare i say decentralisation) etc and "layer" v. lawyers...
* can we evolve the **network layer** (forwarders) to enable better performance, reliability in an application-agnostic way so that PEPs would be less needed?
* What would be useful to say about packets to help networks achieve better performance? (permission-based, with appropriate preferences for treatment, not descriptive of what the packet contains)

2. accountability, privacy, key management and apps that ignore us
* (mpesa, bitcoin etc) because we failed - what do we do now?
* "security" - what do we mean by that, precisely?
-- confidentiality works, mostly; what else?
-- identifier ownership (e.g., sovereign identity)
-- delegation
-- permissions (L3, L7)
-- permission-based sending
-- what can we promise and what is fantasy?
* multiparty trust/confidentiality/integrity/ (perfect fwd secrey without full rekeying etc etc?)
* DDoS resistant architecture
* usage/billing/service differentation?

3. impact of new better computer science (software containers/enclosure/verification - e.g. as used in SDN controller/P4 and eBPF worlds) what does that do to future protocols/systems?

4. APIs & user control (dku)
* what can be learned from today's application needs?
* rethink networking from application perspective (e.g., ICN-style)?

5. Politics, economics, governance, etc?

6. Binary - can we stop talking about userspace and kernel, or fast path and slow, or data and control, or even ascii v. binary coded protocols - how about we comprehend diversity (accelerators, even just multicore esp. numa, means the protocol implementation space is much more "interesting") and also data driven control actions (failure detectors for example, but also triggering route recompute or discovery etc) essentially, the old seperations of concerns are no longer valid, but what replaces them?

7. IPv6 adoption case study?
- Will IPv6 be extensible in the next IPv6 hardware? (is it now/soon possible?)

8. What were the factors that had to be in alignment for the Internet to work? How do we get back to that? What does it mean for the next thing?

9. Open source vs. open standards

* do we need standardized protocol specs in the future?
* You download your "client code" from the web server
* "infection vector" risk?
* copyright and patent discussion as well?

10. Meaning of "end-to-end" today (dku)
* what are the ends in the Internet today?
* next CDN node vs. actual end
* meaning of end-to-end in data-oriented systems (ICN etc.)?
* how do we identify the endpoint/network path?

11. Future infrastructure (dku)
* research opportunities for supporting networked applications, e.g.,
* distributed consensus
* distributed/federated ML
* scalable multi-destination communication
* media processing, JIT transcoding
* generally: computing in the network?

12. What would we want to do differently? (Jari)