1 Naming for Internet 2005. 9. 12. MMLAB, Seongil Han [email protected].

1

Naming for Internet

2005. 9. 12.

MMLAB, Seongil Han

[email protected]

-2-

Contents Overview

A layered Naming Architecture for the Internet

Intentional Naming System (INS)

Problems

Conclusion

-3-

Overview What is naming??

Name <noun> a word or words by which an

individual person, place or thing is identified and referred to

For what? Data, object, service, and so on

In Internet DNS, IP

-4-

Contents Overview

A layered Naming Architecture for the Internet

Intentional Naming System (INS)

Problems

Conclusion

-5-

Current naming (DNS)

2 reference You should

communicate that host with 66.249.89.99 among several service providing hosts

And in 66.249.*.* network

DNS

www.goolgle.co.kr66.249.89.99

www.google.co.kr

I want to use the search engine of google

Application

Transport

Network

Data link

Physical

Application

Network

-6-

New Architecture Design principle #1

Names should bind protocols only to the relevant aspects of the underlying structure; binding protocols to irrelevant details unnecessarily limits flexibility and functionality

DNS system is violated

-7-

New Architecture Decouple I

The name of data or service The endpoint hosting the data or

service

Decouple II The endpoint that I communicate with Its network location (IP address)

-8-

New Architecture Service identifier

Only represents the service or data (not host / endpoint)

Application

Transport

Network

Data link

Physical

SID

EID

Endpoint identifier Only represents the host (or

endpoint) (not its location)

Two resolution SID → (EID,transport,port) EID → IP addresses

The Naming Layers

User-level descriptors(e.g., search)

App session

App-specific search/lookupreturns SID

Transport

Resolves SID to EIDOpens transport conns

IP

Resolves EID to IP

Bind to EID

Use SID as handle

IP hdr EID TCP SID …IP

Transport

App session

Application

-10-

New Namespace Design principle #2

Names, if they are to be persistent, should not impose arbitrary restrictions on the elements to which they refer

DNS, IP are violated

2 approaches Genre (e.g. URN) Flat namespace

-11-

Flat namespace How implement??

DHT O(log n) resolution time → problem Various solutions

The disadvantage Not pay-for-your-own model Why trust cf) RSP

-12-

Contents Overview

A layered Naming Architecture for the Internet

Intentional Naming System (INS)

Problems

Conclusion

-13-

INS is .. Intentional Naming System

Resource discovery and service location system for dynamic and mobile networks of devices and computers

Key features Focus on ‘what’, not ‘where’ Early, late binding Application-controlled metric support Easy deployment on current internet

-14- service

service

service

client

New Architecture

INR networks(Intentional Naming Router)

INR

Early binding

Intentional namenetwork location

DATA

-15- service

service

service

client

New Architecture

INR

Intentional anycast

Intentional name + data

data

-16- service

service

service

client

New Architecture

INR

Intentional multicast

Intentional name + data

-17- service

service

service

client

New Architecture

INR

Discovering Intentional

name

query

names

Announcing an intentional

name

-18-

New naming Intentional name

Name-specifiers : attribute-value pair Wild-card (*) Range matching Example

[city=washington [building=whitehouse [wing=west]]]

[service=camera [data-type=picture [format=jpg]] [resolution=640x480]]

-19-

Name treeroot

accessibility

public

service

camera

resolution

640X480

data-type

picture

city

washington

building

whitehouse

wing

west

Orthogonal attributes

Name-record

-20-

Problems New architecture

Development and deployment is too difficult

Scalability and transition should be significantly considered

New namespace Flat namespace

Ultimate destination, but serious and many challenges

Security

-21-

Conclusion New paradigm is prepared and

appeared by many humans or organizations nowadays

In internet, the change and challenge about naming are needed Ubiquitous computing, mobility

support, security, and so on We need to consider this sufficiently