I2ComM 2008 Colombia, Cartagena February 22 Next Generation Internet Architectures: Current Status and Challenges Fábio Luciano Verdi University of Campinas.

I2ComM 2008 Colombia, CartagenaFebruary 22

Next Generation Internet Architectures: Current Status

and Challenges

Fábio Luciano Verdi

University of Campinas (Unicamp), Brazil


Agenda• Introduction and Background

– Current status of the Internet– Problems

• Some current proposals– IETF– Content Networks

• Pub/sub approach/paradigm

• Current work of our group: architecture and some results

• The future Internet: desired features• Discussion


Introduction and Background

Some years ago…



But it works!

Today…



• Naming

• Shortage of Addresses

• Security

• Mobility

Why so sad?



NamingDNS

Shortage of IPAddresses

NAT

SecurityIPSec

MobilityMobile IP

E2E principle was broken

InternetOssification



• Novel services:– Multimedia– New types of data: voice– IPTV– QoS

• Mobility more dynamic, new places, maybe everywhere!

• Heterogeneity IPv4 / IPv6• Security can be affected by mobility, different threats• Multihoming: end-host multihoming and AS/ISP

multihoming (DFZ problem)• IPv4 is to run out on 22nd May 2010!!!

Even more problems…



• Start immediately a NGI proposal!

• Take into account the experience and the lessons learned so far

• Although we want to construct a “forever” architecture, this is IMPOSSIBLE: unknown situations

• Try to make it simple!!!

What should be done?



The most basic principle:

IP = identifier + locator

separation between identifier and locator


Introduction and Background • 128 bits namespace (solves the lack of IPv4 addresses)• Solves the IP semantic overload• Enables new functionalities, like mobility, multihoming

and heterogeneous network integration

Identifier

Locator

Application

Dynamic Binding

socket

Application

Locator

socket

LocatorLocatorLocator

Static Binding

TCP/IP ID Layer



• We know it is the main solution!!!! But new challenges appear…

• Identifier-based routing X resolution– Where the resolution is done

• Several approaches• Common assumptions: most of the approaches

consider to have a “box” in the border of the ISPs/domains

• Host-based X network-based



ITR ETR

• Mapping EIDs to LOC• Cache• Query

end hostend host









IETF

• LISP (NERD, CONS)

• IPvLX (draft): IPv6 acts as identifiers and IPv4 acts as locators

• eFIT: Separate provider addresses from client addresses (draft, paper at ACM IPv6-August 2007)

IETF proposals (or based on)


Others

• ROFL (Sigcomm 2006)

• UIP (hotnets 2003, 2004, others)

• DONA

• Our nodeID extended:– NID / DID approach


Some Current Proposals: Limitations

• Updates to the mapping are intended to be relatively rare

• Not indicated for fast mobility• Mapping at the edges• Involves an ambitious Replicator system• Cache invalidation• Flooding/distributed servers• Time for lookups









Content Networks• It is a new paradigm!• Content-centric / data-oriented paradigm

– Publish/Subscribe communication model– Information is indexed by keys

and retrieved by subscription.

• Protocols are declarative – Say what you want, not where to get it from

• Data is self-certified – Self-validating data (hash, signature, PKI)– Secure the data not the channel

• Routers/nodes become network processors– Are caches of content, indexes, and buffers.– Forward information while caching,

in the style of MANETS, DTNs, sensor and P2P.


Traditional Internet vs. Content-Centric New Internet


Pub/Sub Communication Model

P

SPublisher Rendevouz Subscription Routing

Dissemination of Publications

ContentAdvertisements

Publisher

Subscriber









Current work of our group: architecture and some results

• Functionalities– Name Resolution– Mobility– Multihoming– Flat Routing– Security– Heterogeneity– Legacy Applications Support


Internet Model

I2ComM 2008 Colombia, CartagenaFebruary 22Network

Mobility

RVS ClientGw Msg SV

DHT Client

DHCP Client

Control plane

Security Mgr

NIDRouting

ConnMapper

Flat Routing Support

NID Filter

DNSHandler

Legacy Appl. Support

SecurityPacketHandler

NIDMapper

Identity Layer

ASI

ANIANI

ARI

Internal Modules

DHCPDNSDHT RVS

External Modules


Results

• Name Resolution (DNS and TXT records)• Registration (RVS and DHT)• Data transfer• Intra-domain mobility• Inter-domain mobility (node and domains)• Multihoming• Heterogeneous networks (IPv4/IPv6)• Network Composition


Results: Composition









The future Internet: desired features

• Jon Crowcroft:

Toward a Network Architecture that does Everything

Communications of the ACM, January 2008



• Unbind identity and location

• Flat and cryptographic global identifier

• Flat routing

• Heterogeneous network integration

• Support to legacy applications (transparency)



• Clean Slate X Patching

• More business-oriented

• Focus on data (content-based routing)

• Sender-oriented receiver-oriented

• Natural or built-in mechanisms for multihoming, security and mobility

• Context-awareness


Discussion

• Challenges– Deployability– Scalability– Compensation mechanisms– Trust– Reputation– Unwanted traffic– …


Thanks!

http://www.dca.fee.unicamp.br/~verdi

I2ComM 2008 Colombia, Cartagena February 22 Next Generation Internet Architectures: Current Status and Challenges Fábio Luciano Verdi University of Campinas.

Documents

locator slide

brazil slide

background current status

lisp lisp

ipv4 networks

agenda introduction

internet transit networks

ipv6 addresses