Transcript poster - Wolfgang Mühlbauer

Trilogy and 4WARD: Towards a Clean Slate Internet Design
Wolfgang Mühlbauer, Roger Karrer, Peter Feil, Anja Feldmann
Trilogy: Re-architecting the Internet
Clean Slate Internet Design – why?
Trilogy’s concept is single architecture that is flexible to adapt to the different forces and demands in the Internet – i.e. a tussle-aware
architecture.
Social
Technical
Many of the design decisions of the Internet were made 4 decades ago. Since then, the Internet and its use have changed as it has
changed itself society. However, today, the Internet is out of shape.
An ossified architecture
The Internet architecture is the result of an initial careful design based on well-argued principles and a 4 decade evolution to add
additional features. Unfortunately, many of these principles are no longer true (security) or new ones have emerged – technical ones
(mobility) as well as economical ones (net neutrality). One thing the Internet lacks is flexibility to efficiently accommodate the new
requirements.
Legal
Business
Control plane: out of shape
The picture of the Internet hourglass to depict its architecture holds for the data plane. However, it is far from true when we consider the
“control plane”. In fact, there is no such thing as an explicit control plane. And second, control protocols had been squeezed into the
architecture to fit the layering structure. Therefore, the Internet architecture has significantly extended its waist over the past 40 years.
It identifies 3 fundamental demands that must be considered jointly in the design of a future Internet architecture
•Reachability Mechanisms
reachability
•Resource Control
mechanisms
•Business Aspects
topology discovery,
reachability
load-dependent,
multi-path
Control plane
Data plane
HTTP
SMTP
NTP
POP3
Telnet
IMAP
FTP
congestion
control
XMPP
SIP
IRC
SDP
ASN-1
RPC
XML
ASP
TCP
RTCP
SCTP
HIP
SSL
RTSP
TLS
Radius
Routing
Multicast
RIP OSPF NHRPP E-IGRP GRE
PIM SM-DM DVMRP
IKE
IPSec
IGP
BGP VRRP EGP GGP IGRP
COPS
MBGP MLD IGMP
CIP
MPLS
RSVP
NSIS
IP Support
802.X
802.3 802.11 802.16
MAC
ATM
802. 802. 802.
SDH PDH xDSL ISDN
3x 11x 16x
FDDI
business
Diameter
MIP
IPv4
routing policy
economic drivers
re-feedback
Trilogy is the first project that tries to unify all of them in a clean slate approach
SMB
RTP
UDP
TRILOGY
resource
control
H.323
DNS
XDR
traffic
engineering
ICMP
ARP
DHCP BOOTP
RARP IARP
Workpackage 1: Reachability
Establish and control transparent reachability in a scalable, dynamic and resilient manner:
•Routing fragility
provide
fast converg•Growing organizational complexity
reachence
ability
Topics include
security
•Routing
•Multi-homing
scalability
Requirements for a
•Remote traffic filtering
provide
identifiers
multiple path
exposure
Routing Architecture
802.1x 802.11i 802.21 802.1D 802.1Q
support for
policies
robust-ness/
stability
Token
Ring
A clean slate Internet design that searches for novel concepts to build an Internet architecture from scratch is needed to ensure
that the Internet is able to address the challenges in the near future.
We at Deutsche Telekom are part of 2 EU FP-7 projects that focus on these challenges:
•4WARD
•Triology
Even though the two projects aims at similar goals, they take different approaches and focus on distinct challenges.
4WARD
Workpackage 2: Resource control
Develop and evaluate a unified approach to resource control that is efficient, fair and incentive-compatible:
•Utilization
•Different fairness metrics
•Cheat-proof
•Congestion
•Storage, battery life, spectrum
ARD
4W
4WARD is based on the following technical premises:
1. Create a new “network of information”, where information
objects have their own identity.
2. Rich communication paths that include mobility, security and
QoS over wired and wireless networks
3. Devise an embedded default-on management capability.
4. Provide means to support the instantiation and dependable
operation of different networks on a single infrastructure in a
commercial setting (virtulization).
5. Develop an integrated framework to represent, design,
implement and operate network architectures that all belong to
a common family of interoperable network instances.
easy
deployability
locality of
routing
events
easy
configuration
DPI;
TE
Resource
user
Trilogy:
Design for
Tussle
Multi-flows;
peer2peer;
DDoS
Resource
provider
Substrate
Workpackage 3: Social and commercial control (led by DT)
Virtualisation of Resources (Task 3.2)
(partitioning of physical infrastructure into “slices”)
Virtualised
substrate
Understand how architectural features allow a controlled behavior, provide flexibility in terms of technical, social and economical
outcomes, and interact with business stakeholders outside Triology to get commercial/strategic steer.
Stakeholder
behaviours
Provisioning of Virtual Networks (Task 3.3)
(aggregate “slices” and form virtual networks)
User
Requirements
System
Requirements
Virtual Network
Virtual Network
Management of Virtual Networks (Task 3.3)
Economic
reality
Market Evolution
Validation
Traditional
V-model
Architecture
and Design
Implementation
System
Testing
(one variant)
Verification
Unit Testing
merge/
network code
decode
Conclusions
split/
balance
code
cooperatively
join
Generic path 1
Generic path 2
A clean slate Internet design
•Requires novel concepts with an out-of-box thinking
•Needs an experimental facility to experimentally verify that protocols and architectures are viable, scalable and efficient
• Offers new technical and economical opportunities for operators, but only for early adopters
For these reasons, DT is involved in the above projects
For further information
•Anja Feldmann: Internet Clean Slate Design: what and why? Sigcomm CCR, July 2007
•For more information on the proposals, please contact any of the authors at [email protected].