Introduction (ppt) - From PURSUIT

parsimoniousknotNetworking and Communications

Feb 16, 2014 (3 years and 1 month ago)

57 views

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

1

Introduction


Arto Karila

Aalto
-
HIIT

arto.karila@hiit.fi

T
-
110.6120


Special Course on Data
Communications Software:
Publish/Subscribe Internetworking


2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

2

ICT and productivity


It is generally believed that increasing use of ICT is the
most important single tool for increasing productivity
(see e.g. OECD study “ICT and Economic Growth…”)


Typically deployment of ICT has increased productivity
by 10 to 20 %, especially when processed have been
revised at the same time


With mobile solutions even 40% increases have been
achieved


Experience from developing countries shows that ICT
can boost productivity there at least as much as in
developed countries


We are still probably utilizing less than 10% of the
opportunities of ICT

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

3

Opportunities of ICT…

ICT can be utilized a lot more in all areas of life:


Public sector:


Health and elderly care


Education


All public services


True openness and direct participation


Enterprises:


Logistics, ERP, CRM, groupware, …


Mobile access to business
-
critical systems


Networking with partners, customers, and others


Integration of voice and video

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

4

… Opportunities of ICT


Private life:


Social media


Entertainment (TV, music, gaming etc.)


Secure and mobile access to public and private
services:


Health, social services, taxes etc.


Education


Banking


etc.

All this requires a lot from

the underlying network!

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

5

Computer networking


Computer networking was developed for mainframes

(on the left ENIAC and on the right IBM S/360)


Sharing devices: computers, mass memory, printers etc.

which have addresses


Traffic is point
-
to
-
point

between two devices

or network interfaces


The old paradigm still

lives even though the

world around has

completely changed


Something has to be

done about this

Picture source: IDG News Service

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

6

History of the Internet…

1957:

Advanced Research Projects Agency (ARPA)


was founded after the launch of the Soviet



satellite Sputnik

1968:

ARPA started the development of the




ARPANET

1969:

The first four nodes of the ARPANET were



connected (the first message: ”lo”)

1974:

Robert Kahn and Vinton Cerf defined the basic


Internet architecture (TCP/IP)

1975:

DARPA started the development of Internet



technology

1983:

On 1/1/1983 the ARPANET was converted to TCP/IP

BSD 4.2 had TCP/IP protocol stack

1988:

FUNET joined the Internet

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

7

… History of the Internet

1989:

Telecom Finland published DataNet

BGP
-
1 was defined

1990:

NSFNET was founded

1991:

The first World Wide Web (WWW) client Mosaic
was published at CERN

1993:

CIDR and BGP
-
4 were adopted

1990’s:

The Internet secured its position as the leading
network architecture

2000:

The number of Internet hosts exceeded
100,000,000

2011:

The number of Internet hosts is approaching

1 billion (1000,000,000)

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

8

Growth of the Internet

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

9

Problems with the current Internet


Over the past 30 years, several major changes have
been made to the Internet


always at the last moment


Internet’s success is largely based on its ability to
adapt to the changing requirements


With these changes, the end
-
to
-
end principle is already
destroyed by middle
-
boxes (NAT and firewalls)


We have reached a point, where the Internet is ossified
and new transport protocols are virtually impossible


The Internet should be able to accommodate a wider
range of tussles


We need a clear separation of the naming space and
network functions


The Internet is working on the terms of the sender

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

10

Why the Internet only just works


See:
Why the Internet only just works
, M. Handley,

BT Technology Journal, Vol 24 No 3, July 2006


Throughout its life, the Internet has only just worked
and all of the major changes have been made at the
last possible time


CIDR and NAT were introduced because of the
exhaustion of the IPv4 address space


These were supposed to be temporary solutions,
waiting for IPv6 to break through, but they have
become permanent


At the same time firewalls proliferated


The end
-
to
-
end principle of the Internet no longer
works because of the middle boxes (firewalls and NAT)


This has lead to it being virtually impossible to make
any changes to the transport layer (TCP/UDP)

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

11

Ossification of the Internet


We have ended up in a vicious circle:


Developers cannot use a new protocol because it
cannot traverse firewalls and NAT


It is not worth while for the developers of firewalls
and NAT to change the middle boxes because there
are no users of new transport protocols


No major changes have been made to the core
protocols of the Internet since 1993


The core protocols of the Internet are ossified while
the needs have developed significantly


Innovation in the Internet is withering

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

12

Security and Trust


Junk mail (Spam) and other types of unsolicited traffic
are growing problems


There still are no effective defense strategies against
Distributed Denial of Service (DDoS) attacks


Worms, viruses, Trojan horses, spyware, and other
malware is spreading fast throughout the Internet


Phishing is a growing problem


The Internet was developed for a community where
everybody was assumed trustworthy


now trust in the
Internet has eroded


Now that the Internet is used by everybody, we need to
enable communication between distrusting parties


We need mechanisms by which people and companies
can build and evaluate trust


Combining privacy and reputation is challenging

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

13

Some challenges of the Internet


Among the well understood requirements for the
Internet are the following:


Multicast


Mobility


Multi
-
homing


Security


Quality of Service (QoS)


Ability to handle massive video (including IPTV)


Scalability to future needs (Network of Things etc.)


Solutions to many of the needs listed above have been
developed but not widely deployed


Operators don’t have incentives to bring new features
to the market because they are only useful if they are
interoperable with other operators, in which case they
give no competitive advantage

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

14

IPv6


IPv6 was defined in 1995 and expected to spread fast


It is still hardly used in Western countries


The main improvement of IPv6 is moving from 32
-
bit to
128
-
bit addresses


IPv6 was defined at a time when nobody could foresee
all of the uses and needs of the Internet that we have
now


CIDR and NAT have eased the shortage of IPv4
addresses but now they are really running out


The transition to IPv6 will be a long one and it won’t
solve most of the problems

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

15

Expensive transit


The current inter
-
operator routing protocol BGP
-
4

does not fulfill modern requirements but there is

no successor to it in sight


Tier
-
1 operators (AT&T, MCI, Sprint, C&W etc.) are a
group of about a dozen global operators with mutual
peering agreements


Tier
-
1 operators don’t pay for transit while others pay
to them (tier
-
2 operators directly and others indirectly)


In Practice they form a cartel, which wants to cement
the market and is not advocating development

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

16

Storage vs. Transit Price

1985

1990

1995

2000

2005

2010

$100/MB

$10/MB

$1/MB

$100/GB

$10/GB

$1/GB

$0.1/GB

2009

Source: Dr. Pekka Nikander

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

17

Microeconomics


Over the past decade, microeconomics have grown in
importance


We need economic mechanisms that encourage
people to do good for the community


The Internet was developed with public funds for
research and education without any commercial
considerations


If we want to inject resources into the network, it
must be possible for the party paying for them to also
receive (some of) the revenues


We need to create ways for companies and people to
improve their own economies by doing things
beneficial for the community

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

18

Users


Applications


Data


Users, applications and data are involved in computing


All three are becoming increasingly mobile


The network has to bring these three together in a
reliable, secure and efficient way

Network

Users

Appl.

Data

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

19

Clouds and Grids


Applications are increasingly run in cloud and grid
environments


Cloud computing was created to cut down the cost and
increase the flexibility of computing


In a cloud, dynamically scalable and often virtualized
ICT resources are offered as services over the Internet


Google started packing cheap off
-
the
-
shelf computers
and DC UPS’s into containers and placing them every
-

where, cutting the cost of data centers by a factor of 10


While clouds still are based on computing centers,
grids can run in millions of PCs


With ever more powerful portable devices and the
proliferation of mobile data, also grids will be
increasingly mobile

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

20

Scalability


The Internet has already scaled to a level that was
unconceivable to its original developers


However, new trends will raise the scalability
requirement of the Internet to a much higher level:


Proliferation of video (YouTube, IPTV etc.)


Ubiquitous computing

(
ubicomp
)


Sensor networks


Internet of things


The amount of video traffic is growing rapidly in
wireline and wireless networks


We have to be ready for dozens and hundreds of
billions of nodes in the network in the near future


The capacities and abilities of nodes will vary highly

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

21

About applications


Most applications are still generic in nature and
basically the same as in the 1980’s (e.g. office suites)


On the other hand, ERP systems (such as SAP) tend to
cement the existing flawed processes


We should be developing applications that directly
support work flows thereby increasing productivity


With modern tools (e.g. AJAX and QT) and methods
(e.g. agile programming) we should be able to cut
down the development time and cost by a factor of 10


Middleware is getting standardized and applications
becoming component
-
based, easing integration


Applications are dealing with information, which is
structured and linked

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

22

Requirements


The network has to meet the needs of the applications
of today and the future:


Mobility of users, data and computation


Scalability up to hundreds of billions of nodes


Efficient handling of video


In
-
built security, including protection against SPAM
and DoS attacks


We are interested in information content


not who is
storing it and where


Network has to support access to and processing of
large amounts of hierarchically organized information


There needs to be a simple, powerful and efficient API
for accessing the services of the network


the API
could be generic and run on different networks

2011
-
09
-
12 AK

T
-
110.6120: Publish/Subscribe Internetworking

23

Conclusions


There is a growing consensus among researchers of
internetworking that a fundamental reform is needed


We need to be able to name and address information
rather than hosts or interfaces


We need mechanisms for structuring information and
limiting its visibility


We need to have a way to store information graphs in
the network and retrieve and process them in an
efficient way, not caring about their whereabouts


Information Centric Networking (ICN) and, more
specifically, the Publish/Subscribe (pub/sub) paradigm
seem to offer solutions to our needs


PSIRP/PURSUIT is an attempt to that direction