adding interlligence to internet

8/12/2019 Adding Interlligence to Internet

1/31

Seminar Report 03 Adding Intelligence to Internet

1. INTRODUCTION

Satellites have been used for years to provide communication network

links. Historically, the use of satellites in the Internet can be divided into two

generations. In the first generation, satellites were simply used to provide

commodity links (e.g., T1 between countries. Internet !rotocol (I! routers

were attached to the link endpoints to use the links as single"hop alternatives to

multiple terrestrial hops. Two characteristics marked these first"generation

systems# they had limited bandwidth, and they had large latencies that were due

to the propagation delay to the high orbit position of a geosynchronous satellite.

In the second generation of systems now appearing, intelligenceis added at the

satellite link endpoints to overcome these characteristics. This intelligence is

used as the basis for a system for providing Internet access engineered using a

collection orfleetof satellites, rather than operating single satellite channels in

isolation. $%amples of intelligent control of a fleet include monitoring whichdocuments are delivered over the system to make decisions adaptively on how

to schedule satellite time& dynamically creating multicast groups based on

monitored data to conserve satellite bandwidth& caching documents at all

satellite channel endpoints& and anticipating user demands to hide latency.

This paper e%amines several key 'uestions arising in the design of a

satellite"based system#

an international Internet access using a geosynchronous satellite be

competitive with today)s terrestrial networks*

+hat elements constitute an intelligent control for a satellite"based

Internet link*

+hat are the design issues that are critical to the efficient use of satellite

channels*

Dept. of IT MESCE, Kuttippuram1


2/31


The paper is organi-ed as follows. The ne%t section, Section ,

e%amines the above 'uestions in enumerating principles for second"generation

satellite delivery systems. Section / presents a case study of the Internet

0elivery System (I0S, which is currently undergoing worldwide field trials.

Dept. of IT MESCE, Kuttippuram


3/31


2. ISSUES IN SECOND-GENERATION SATELLITE LINK

CONTROL

+e discuss in this section each of the 'uestions raised in this paper)s

introduction.

Can international Internet access using a geosnc!ronous satellite "e

co#$etiti%e &it! to'a(s terrestrial net&or)s*

The first 'uestion is whether it makes sense today to use

geosynchronous satellite links for Internet access. lternatives include wired

terrestrial connections, low earth orbiting (2$3 satellites, and wireless wide

area network technologies (such as 2ocal 4ultipoint 0istribution Service or

.5"6H- radio links in the 7.S..

+e see three reasons why geosynchronous satellites will be used forsome years to come for internationalInternet connections. The first reason is

obvious# it will be years before terrestrial networks are able to provide ade'uate

bandwidth uniformly around the world, given the e%plosive growth in Internet

bandwidth demand and the amount of the world that is still unwired.

6eosynchronous satellites can provide immediate relief. They can improve

service to bandwidth"starved regions of the globe where terrestrial networks are

insufficient and can supplement terrestrial networks elsewhere.

Second, geosynchronous satellites allow direct single"hop access to

the Internet backbone, bypassing congestion points and providing faster access

time and higher net throughputs. In theory, a bit can be sent the distance of an

international connection over fiber in a time on the order of tens of

microseconds. In practice today, however, international connections via

Dept. of IT MESCE, Kuttippuram/


4/31


terrestrial links are an order of magnitude larger. 8or e%ample, in e%periments

we performed in 0ecember 199:, the mean round trip time between the 7.S.

and ;ra-il (vt.edu to embr.net.br over terrestrial links were .?. In contrast, the mean latency between the two routers

at the two endpoints of a satellite link between ;angladesh and Singapore

measured in 8ebruary 1999 was /5:.< msec. Therefore, a geosynchronous

satellite has a sufficiently large footprint over the earth that it can be used to

create wormholesin the Internet# constant"latency transit paths between distant

points on the globe @henA. The mean latency of an international connection via

satellite is competitive with today)s terrestrial"based connections, but the

variance in latency can be reduced.

s 'uality"of"service (BoS guarantees are introduced by carriers, the

mean and variance in latency should go down for international connections,

reducing the appeal of geosynchronous satellites. However, although BoS may

soon be widely available within certain countries, it may be some time until it is

available at low cost between most countries of the world.

third reason for using geosynchronous satellites is that the Internet)s

traffic distribution is not uniform worldwide# clients in all countries of the world

access content (e.g., +eb pages, streaming media that today is chiefly produced

in a few regions of the world (e.g., Corth merica. This implies that a

worldwide multicast architecture that caches content on both edges of the

satellite network (i.e., near the content providers as well as near the clients

could provide improved response time to clients worldwide. +e use this traffic

pattern in the system described in the case study (Section /.

3ne final point of interest is to ask whether 2$3 satellites that are

being deployed today will displace the need for geosynchronous satellites. The

low orbital position makes the 2$3 footprint relatively small. Therefore,

http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#ref_Chenhttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#ref_Chen


5/31


international connections through 2$3s will re'uire multiple hops in space,

much as today)s satellite"based wireless phone systems operate. The

propagation delay will eliminate any advantage that 2$3s have over

geosynchronous satellites. 3n the other hand, 2$3s have an advantage# they

are not subDect to the constraint in orbital positions facing geosynchronous

satellite operators. So the total available 2$3 bandwidth could one day surpass

that of geosynchronous satellites.

+!at ele#ents constitute an ,intelligent control, or a satellite-"ase'

Internet lin)*

The basic architecture behind intelligent control for a satellite fleet is

to augment the routers at each end of a satellite link with a bank of network"

attached servers that implement algorithms appropriate for the types of traffic

carried over the links. +e use certain terminology in our discussion. 8irst, given

the argument above for asymmetric traffic, our discussion is framed in terms of

connecting content providers(in a few countries to end users (in all countries.

In some cases (e.g., two"way audio, however, the traffic may be symmetrical.

Second, we refer to the content"provider endpoint of a satellite link as a

warehouse, and the end"user endpoint as a kiosk. The architecture of

warehouses and kiosks must be scalable# The number of servers, storage

capacity, and throughput of warehouses and kiosks must scale as the number

and bandwidth of satellite links, content providers, and end users grows.

8igure 1 illustrates the generic architecture. ontent providers are

connected via the terrestrial Internet to a router inside a warehouse. The router

also connects to a local area network that interconnects various servers. The

router also connects to the earth station for the satellite. +ithin the footprint of

the satellite are many groundstations, each connected to a router within a kiosk.

The kiosk is similar to the warehouse in that it connects to a local area network



6/31


that interconnects servers, and optionally, to a terrestrial Internet connection.

The kiosk also acts as the head end for Internet service providers (IS!s that

provide network connections to end users. 4ore details are given in the case

study in Section /.

igure 1/ Intelligent control resi'es in &are!ouses an' )ios)s

Intelligent controls reside in the warehouse and kiosk and are re'uired

to share limited satellite bandwidth among many users and to hide the 'uarter"

second latency of a geosynchronous satellite. The controls are a distributed

algorithm, in which part runs on warehouses and part runs on kiosks. ll

warehouses and kiosks must cooperate and must coordinate the use of satellite

resources. 4ulticast groups are defined to allow communication between

cooperating entities (e.g., between a warehouse and multiple kiosks.

To identify which controls make sense, it is useful to look at the

characteristics of Internet traffic. 8igure is a ta%onomy of traffic with si%

categories. Three of them represent +eb pages# pages that are popular for

Dept. of IT MESCE, Kuttippuram=
http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#fig_traffichttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#fig_traffic


7/31


months or longer (e.g., a news service such as cnn.com& pages that are popular

for a short time (e.g., hours, days, or weeks, such as those resulting from

3lympic games& and pages that are accessed only a few times. 3ne of the facts

known about this traffic is that most of the re'uests and most of the bytes

transferred in client workloads come from a small number of servers. 8or

e%ample, in a study of pro%y or client uniform resource locator (7E2 reference

traces from 0igital $'uipment orporation (0$, merica 3nline, ;oston

7niversity, Firginia Tech, a gateway to South Gorea, and one high school, :>

to 9


8/31


aching of both categories of popular 7E2s and push channels should be

done at both the warehouse and the kiosk. aching at the kiosk side

obviously avoids the satellite delay when an end"user re'uests a popular

document. aching at the warehouse is desirable to decouple the process

of retrieving documents from the content providers (i.e., the path

between content providers and the warehouse in 8igure 1 and the

process of scheduling multicast transmission of documents from a

warehouse via satellite to kiosks. In addition, the warehouse reduces

cache consistency traffic, because consistency traffic occurs only

between the content providers and the warehouse. The kiosks do not need

consistency checks, because they rely on the warehouse to send them

updated pages when the warehouse detects inconsistency.

8eedback of logs of documents, channel content, real"time streams, and

timely documents are re'uested from the kiosk end of satellite

connections by the warehouse side for use in adaptive algorithms.

7npopular pages may be cached at an individual kiosk only, and

retrieved from the Internet using a terrestrial link if available. 3nly if

feedback from the kiosk logs sent to the warehouse shows that a

document is popular among multiple kiosks does the document get

reclassified as popular for short term and hence cached at the

warehouse.

To hide latency, pages and updates to the content of changed pages could

be preemptively delivered across the satellite link based on the feedback

of logs.

!ush channels could be dynamically constructed by identifying which

+eb documents have become popular.

4ulticast could be used to deliver push channels, real"time streams and

timely documents based on subscriptions.

Dept. of IT MESCE, Kuttippuram:


9/31


The bandwidth available on satellite links could be allocated based on

traffic categories.

+!at are t!e 'esign issues t!at are critical to using satellite c!annels

eicientl*

The overall system must achieve a balance between the throughput of

the terrestrial Internet connection going into the warehouse, the throughput of

the warehouse itself, the throughput of the satellite link, the throughput of each

kiosk, and the throughput of the connection between a kiosk and its end users.

In addition, a balance among the number of end users, the number of kiosks,

and the number of warehouses is re'uired.

onsider some e%amples. s the number of end users grows, so will

the si-e of the set of popular +eb pages that must be delivered, and the

bandwidth re'uired for push, real time, and timely traffic. 2et)s look at +eb

traffic in detail. nalysis of end"user traffic to pro%y servers at merica 3nline

done at Firginia Tech shows that an average user re'uests one 7E2 about every

seconds, which indicates a re'uest rate of >.> 7E2s per second. (This does

not mean that a person clicks on a link or types a new 7E2 every seconds&

instead, each 7E2 re'uested typically embeds other 7E2s, such as images. The

average rate of the individual 7E2s re'uested either by a person or indirectly as

an embedded obDect is one every seconds. Thus, a kiosk supporting 1,>>>

concurrentusers must handle a re'uest rate of >> per second. The median filesi-e from the set of traces cited above (0$, merica 3nline, etc. is

kilobytes. Thus, the kiosk Hyperte%t Transfer !rotocol (HTT!"level throughput

to end users must be 5>> kilobytes per second. t the other end, the warehouse

has a connection to the Internet. The bandwidth of this connection must e%ceed

that of the satellite connection, because the warehouse generates cache

consistency traffic. The servers within the warehouse and kiosk have limited

throughput, for e%ample, the throughput at which the cache engines can serve



10/31


+eb pages. To do multicast transmission, a collection of content (+eb pages,

pushed documents must be bundled up at the application layer at the warehouse

into a unit for transmission to a multicast group, then broken down into

individual obDects at the kiosk. This assembly and disassembly process also

limits throughput.

second issue is how to handle +eb page misses as kiosks. If the

kiosk has no terrestrial Internet connection, then these situations obviously must

be satisfied over the satellite channel. This reduces the number of kiosks that a

satellite link can handle. 3n the other hand, if the kiosk does have a terrestrial

connection, an adaptive decision might be to choose the satellite over the

terrestrial link if there is unused satellite capacity and if the performance of the

territorial link is erratic.

third issue is how to handle 0omain Came System (0CS lookups.

0CS server is necessary at kiosks to avoid the delay of sending lookups over

a satellite. However, how should misses or lookups of invalidated entries in the

kiosk)s 0CS server be handled* 3ne option is for the 0CS traffic to go over a

terrestrial link at the kiosk, if one is available. n alternative is for the

warehouse to multicast 0CS entries to the kiosks, based on host names

encountered in the logs transmitted from the kiosks to the warehouse.

fourth issue is fault tolerance. If a kiosk goes down and reboots, or

a new kiosk is brought up, there must be a mechanism for that kiosk to obtain

information missed during the failure.

Dept. of IT MESCE, Kuttippuram1>


11/31


. CASE STUD/ INTERNET DELI3ER SSTE4

IDS uses

multicast transmission to share channel bandwidth with users in many

counties

caching (e.g., Table 1 at both ends of the satellite to hide or avoid

latency, in the form of large (terabyte"si-e content warehouses and

kiosks

automated monitoring of user behavior to dynamically create multicast

push channels of content

proactive content refreshing that updates inconsistent cached documents

before users re'uest those documents

The obDective of the I0S is to provide fast and economical Internet

connectivity worldwide. I0S also facilitates Internet access to parts of the globethat have poor terrestrial connectivity. I0S achieves this goal by two means#

1. creating satellite"based wormholes@henA, from content providers to

geographically distant service providers, thus providing a fast path

from one edge of the network to the other

. caching content such as HTT!, 8ile Transfer !rotocol (8T!, CCT!,

and streaming media at the content"provider end as well as the

service"provider end, thus conserving bandwidth

The idea for the I0S was conceived at ICT$2ST, an international

organi-ation that owns a fleet of geostationary satellites and sells space segment

bandwidth to its international signatories. +ork on the prototype started in

8ebruary 199:. In 8ebruary 1999, the prototype system stands poised for

international trials involving ten signatories of ICT$2ST. commercial

version of I0S will be released in 4ay 1999.

http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#ref_Chenhttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#ref_Chen


12/31


The building blocks of I0S are warehousesand kiosks. warehouse

is a large repository (terabytes of storage of +eb content. The warehouse is

connected to the content"provider edge of the Internet by a high"bandwidth link.

6iven the global distribution of +eb content today, an e%cellent choice for a

warehouse could be a large data"center or large"scale bandwidth reseller situated

in the 7.S. The warehouse will use its high"bandwidth link to the content

providers to crawl and gather +eb content of interest in its +eb cache. The

warehouse uses an adaptive refreshing techni'ue to assure the freshness of the

content stored in its +eb cache. The +eb content stored in the warehouse cache

is continuously scheduled for transmission via a satellite and multicast to a

group of kiosks that subscribe to the warehouse.

The centerpiece of the kiosk architecture is also a +eb cache. Giosks

represent the service"provider edge of the Internet and can therefore reside at

national service providers or IS!s. The storage si-e of a kiosk cache can

therefore vary from a low number of gigabytes to terabytes. +eb content

multicast by the warehouse is received, is filtered for subscription, and is

subse'uently pushed in the kiosk cache. The kiosk +eb cache also operates in

the traditional pull mode. ll user re'uests for +eb content to the service

provider are transparently intercepted and redirected to the kiosk +eb cache.

The cache serves the user re'uest directly if it has the re'uested content&

otherwise, it uses its link to the Internet to retrieve the content from the origin

+eb site. The cache stores a copy of the re'uested content while passing it back

to the user who re'uested it.

The layout for an I0S prototype warehouse and kiosk is shown in

8igure 1 . The prototype warehouse consists of two server class !entium II

based machines, namely an application serverand a cache server. The cache

server houses a +eb cache and other related modules. The +eb cache at the

http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#fig1http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#fig1


13/31


warehouse has 1>> gigabytes of storage. The application server is host to a

transmitter application, a relational database, and a ava"based management

application. These servers reside on a dedicated subnet of the warehouse

network. This subnet is connected to a multicast"enabled router that routes all

multicast traffic to a serial interface for uplinking to the ICT$2ST I0E system

@IntelsatA. The ICT$2ST I0E system provides I! connectivity, over a "4bps

satellite channel, between the warehouse and kiosks.

The prototype kiosk also contains a !entium II"based application

serverand a !entium II"based cache server. The kiosk cache server houses a

+eb cache with gigabytes of storage. The application server is host to a

receiver application, a relational database, and a ava"based configuration and

management application. These servers reside on a dedicated subnet of the kiosk

network. This subnet is connected to a multicast"enabled router. n important

part of the prototype kiosk is a layer"5 server switch @+illiamsA, which is used

to transparently redirect all HTT! (Transmission ontrol !rotocolJport :> user

traffic to the kiosk cache server.

I0S treats +eb content as composed of si% traffic categories as

categori-ed in 8igure . These categories are summari-ed in Table 1 below.

Type traffic consists of HTT! +eb content that is identified by a human

operator as content that should remain popular over a long time (e.g., months.

This may include popular news +eb sites such as the able Cews Cetwork

(CC +eb site. Type ; traffic refers to HTT! +eb content directly pushed into

the warehouse by subscribing content providers. Type $ traffic refers to unicast

HTT! user re'uest"reply traffic that passes though the kiosk and is not cached at

the kiosk. The reply for a type $ re'uest is cached at the kiosk on its return path

from the origin server. s re'uests for a particular 7E2 accumulate at multiple

kiosks, such a hot"spot 7E2 is converted from type $ to type . Type 0 traffic

refers to real"time streaming traffic. Type 8 traffic refers to semi"real"time

Dept. of IT MESCE, Kuttippuram1/
http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#ref_intelsathttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#williamshttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#fig_traffichttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#table1http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#ref_intelsathttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#williamshttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#fig_traffichttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#table1


14/31


reliable traffic such as financial 'uotes and CCT!. Traffic of types , ;, , 0,

or 8 is multicast to all kiosks and pushed to subscribing kiosks.

Ta"le 1/ IDS traic categories

Traic Deinition IDS traic categor

+eb sites popular over long time

!ublisher pushed ;

utomatically identified as hot +eb sites

Giosk unicast re'uests $

Eeal"time streaming 0

Semi"real time 8

The I0S prototype implements traffic types , , and $. 8igure /

below shows the flow of traffic types , , and $ through the I0S system. Type

traffic is defined by the warehouse operator by entering popular 7E2s

through the warehouse management interface. The warehouse operator also

classifies 7E2s into channels as part of creating type content. 3nce created,

content belonging to type is registered in the relational database andsubse'uently crawled from the +eb and stored in the warehouse +eb cache.

The warehouse refreshes content of type from the origin servers based on an

adaptive refresh algorithm. ontent of type is also continuously multicast by

the transmitter application to the kiosks. t the kiosk, the receiver application

filters the incoming multicast traffic, thus accepting only the subset of traffic

that belongs to channels subscribed at the kiosk. 8iltered content is then pushed

into the +eb cache at the kiosk.

Traffic of type $ originates as an HTT! re'uest from kiosk end users.

The re'uest is redirected to the layer"5 switch at the kiosk, the kiosk cache. If

the re'uested content is not found in the kiosk cache, then that re'uest is routed

to the origin server. The reply from the origin server is cached at the kiosk +eb

cache on its way back to the end user who made the re'uest. In 8igure , the

http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Fig4http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Fig4


15/31


path for unicast type $ traffic is shown as going through the satellite back

channel. It must be noted, however, that type $ traffic bypasses all warehouse

components and is routed to the Internet.

3n a periodic basis, the warehouse polls all subscribing kiosks for hit

statistics regarding the type $ content in their respective +eb caches. 7sing this

information and appropriate business rules specified by the management

application at the warehouse, the warehouse converts a subset of type $ content

to type . 3nce type content has been created, the data flow for this traffic

type follows the same path as described above for traffic type .

igure / IDS 'ata lo&



16/31


5. IDS ARC6ITECTURE

Co#$onents at t!e &are!ouse

The I0S warehouse is composed of four maDor components, namely

the cache subsystem, transmission subsystem, management subsystem, and

database subsystem. 8igure 5shows the maDor components of the warehouse

along with their interconnections.

The cache subsystem consists of a cluster of standard +eb caches that

communicate among each other using standard protocols such as Internet ache

!rotocol (I!. 8or the I0S prototype, we have a single S'uid cache. The cache

subsystem also consists of refresh and crawl modules that communicate with the

+eb cache(s using HTT! and are responsible for proactively refreshing and

crawling newly created type or content from origin +eb servers,

respectively. The log module in the cache subsystem parses standard logs from

the +eb cache and communicates hit"metering data to the database subsystem.

The transmission subsystem contains scheduling and gathering modules. These

modules perform the following functions#

obtain from the database subsystem lists of 7E2s belonging to content of

types and that must be transmitted

obtain obDects corresponding to the 7E2s from the +eb cache

append a bit map denoting kiosks that subscribe to the channel with each

7E2 to the corresponding obDect

construct obDect bundles si-ed for optimal transmission

transfer the bundles to the transmitter

Dept. of IT MESCE, Kuttippuram1=
http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Fig2http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Fig2


17/31


The transmitter module, also a part of the transmission subsystem,

receives bundles and transmits them using the 4ulticast 8ile Transfer !rotocol

from Starburst ommunications.

The management subsystem is a +eb"based graphical front end that

communicates with the database subsystem and provides the warehouse

operator with a tool to perform the following types of activities#

add popular content to the warehouse (traffic type

create channels and manually classify content into channels

configure system operational parameters such as thresholds for

conversion of content from type $ to type

The database subsystem consists of the relational database, the K

module, and the mapper. The relational database contains persistent information

about the content stored in the warehouse +eb cache as well as 7E2 hit

statistics and channel and subscription information. The K module performs

three maDor tasks#

1. It periodically re'uests per"7E2 hit statistics for type $ content

stored in the +eb caches of all kiosks.

. ;ased on thresholds set at the warehouse, it converts a subset of $

content to .

/. It automatically channeli-es the newly converted content into

the channels available at the warehouse.

Dept. of IT MESCE, Kuttippuram1?


18/31


Co#$onents at t!e )ios)

2ike the warehouse, the I0S kiosk is also composed of four maDor

components# (1 the cache subsystem, ( the transmission subsystem, (/ the

management subsystem, and (5 the database subsystem. 8igure


19/31


The management subsystem at the kiosk is a +eb"based graphical

front end that communicates with the database subsystem and provides the

kiosk operator with a tool to perform the following types of activities#

subscribing to the channels made available by the warehouse

blocking specific 7E2s from being pushed into the kiosk +eb cache

configuring system operational parameters

The database subsystem consists of the relational database and the K

module. The relational database contains persistent information about the

content stored in the kiosk +eb cache as well as 7E2 hit statistics and channel

and subscription information. The K module at the kiosk transmits per"7E2 hit

statistics for $ type content stored in its +eb cache when re'uested to do so by

the warehouse.



20/31


7. DESIGN ISSUES AND GOALS

In this section, we discuss some of the salient design goals that make

I0S a uni'ue system that fits its re'uirements.

Content reres! at t!e &are!ouse

ache consistency techni'ues for +eb caches are a well"debated

topic. In the authors describe techni'ues for maintaining fresh content in +eb

caches, namely,

Time to live (TT2 fields, implemented using the e%pires header field

in HTT!, and are used by content publishers to set a TT2 for obDects they

create. +hen the TT2 for an obDect e%pires, the cache invalidates that

obDect. The ne%t re'uest for that obDect to the +eb cache is directed to the

origin server. The caveat in this strategy is that for a significant

proportion of +eb content, the TT2s are either incorrect or unspecified.

lient polling is a techni'ue in which caches periodically 'uery the

origin server to determine whether an obDect has changed. The 'uery

fre'uency is a key factor in this techni'ue. The authors used a client"

polling scheme based on the assumption that young files are modified

more fre'uently than old files on the le% 8T! cache. well"tuned

client"polling algorithm can be more effective than a cache refresh policy

based only on TT2.

In the I0S design, the warehouse maintains the freshness of obDects

residing within I0S (i.e., in the warehouse and kiosk +eb caches. The I0S

Dept. of IT MESCE, Kuttippuram>


21/31


warehouse design includes an adaptive refresh client"polling techni'ue that uses

obDect TT2s as initial estimates for obDect refresh times. The client"polling

techni'ue is encapsulated in the following relationship#

CiL1M CiLf.(Ci"M

where Cidenotes the time when the ith 'uery to check the freshness

of an obDect was sent to the origin +eb server. CiL1denotes the estimated 'uery

time for the (i + 1th 'uery. M denotes the time when the obDect was last

modified at the server. 8inally,fdenotes a constant factor. desirable value of f

can be determined by minimi-ing the number of 'ueries i, such that a

subse'uent modification to the obDect after timeMis discovered as 'uickly as

possible. value of >.1 for f is suggested in the HTT! 1.1 re'uest for

comments.

;y having the warehouse refresh all the kiosks, I0S saves the client"

polling bandwidth to origin servers. In addition, the refresh mechanism in I0S

is sensitive to obDects that change too fre'uently. Such fre'uently changing

documents are tagged as uncacheable by the system.

Content $reetc!

The I0S warehouse is designed to prefetch cacheable obDects

embedded in a cached +eb page. The crawler module in I0S proactively parses

cached obDects for embedded cacheable obDects, evaluates the embedded obDectsagainst evaluation parameters, and fetches them from their origin servers to be

cached. The evaluation parameters, set through the management application, use

the heuristic that obDects associated with a popular obDect are likely to be

popular also. Thus, caching of prefetched obDects leads to better hit rates for the

kiosk caches.



22/31


Content rerun -- )ios) ault tolerance

long with new and updated +eb content, the I0S warehouse

constantly multicasts all information cached in its +eb cache to the kiosks. This

design feature provides an automatic recovery for kiosks that were offline for a

certain period of time. It automatically brings new kiosks up"to"date as well.

8us! c!annels

The I0S warehouse is designed to classify all +eb content into push

channels based on keywords associated with cached obDects. Two methods of

channeli-ation are present. 4anual channeli-ation is offered through the

management application at the warehouse. ny obDect in the warehouse can be

manually associated with a channel. +eb content brought in by the warehouse

crawler module is also automatically channeli-ed based on a keywords

discovery algorithm in the crawler. Giosks subscribe to channels offered by the

warehouse. ;ased on kiosk subscriptions, which are communicated periodically

from all kiosks to the warehouse, the warehouse is able to append a subscription

bitmap to all obDects being multicast out to the kiosks. Giosks inspect the

subscription bitmap and filter out all unwanted traffic.

8orta"le #o'ule interaces

specific goal in the design of I0S is to design modules with

portable interfaces. This allows fle%ibility in choosing implementation

platforms. ll I0S modules within the warehouse and kiosks use Transmission

ontrol !rotocolJI! for interprocess communication. Thus, all warehouse (or

kiosk functionality may reside on a single machine or may be distributed

among several machines. In addition, the relational database communicates

with other modules through a single set of application programming interfaces.



23/31


Trans$arent re'irection o 6TT8 traic at )ios)

hard re'uirement of the I0S kiosk design is that all HTT! traffic

from users to the kiosk must be transparently redirected to the kiosk cache.

Thus, the deployment of a kiosk at a national service provider or an IS! will be

invisible to the customers of the kiosk. lthough transparent redirection has

been implemented in software by Cetcache and other cache vendors, most

service providers choose to deploy a hardware"based solution such as +eb

ache ontrol !rotocol running on IS3 cache engines and routers or to use a

content"aware layer"5 switch. The I0S design includes a layer"5 switch to

implement transparent redirection of HTT! traffic to the kiosk cache.

Content $us! into +e" cac!e

The +eb caches used in I0S are designed to operate inpullas well as

pushmode. +hile pull is the default mode of behavior, I0S +eb caches are

modified to accept obDect push. The kiosk +eb cache accepts obDects pushed

from the warehouse. The warehouse +eb cache can accept obDects pushed from

content providers.

4ini#al #o'iications to +e" cac!e arc!itecture

key goal in the I0S design was to keep modification to the +eb

cache to a minimum, which would allow the +eb cache to be used as a

pluggable module. The I0S prototype uses the S'uid +eb cache from the

Cational 2aboratory for pplied Cetwork Eesearch& however, the design allows

for the substitution of S'uid by any commercial +eb cache that implements the

push protocol.

Dept. of IT MESCE, Kuttippuram/


24/31


8ersistent storage o +e" cac!e #eta'ata

+eb cache metadata including hit"metering statistics for all cached

obDects is stored in a relational database. This persistent storage provides I0S

with the ability to 'uery relevant metadata statistics to enforce business rules.



25/31


9. ROAD4A8 OR T6E UTURE

lthough the concept of shortcuts from the content providers to

service providers is not new, this idea has been harnessed only recently. This

recent development has resulted from the availability of enabling technologies

such as I! multicastJbroadcast over satellite, +eb caches that accept push, and

transparent redirection of layer"5 traffic. +hile the planet becomes wired

through terrestrial and undersea high"bandwidth fiber links, geostationary

satellites offer an ideal platform for offering an intelligent and high"performance

infrastructure for Internet delivery. The first generation of such intelligent

products will be available this year. These products and services will also serve

as the proving grounds for several ne%t"generation internet services, such as

virtual private networks (F!Cs, which offer security as well as guaranteed

'uality of service from the content providers to the end users.

t the writing of this document the I0S prototype has been tested at

ICT$2ST labs and is poised for deployment in international trials. The trials

will comprise ten signatories of ICT$2ST, including Teleglobe International

(anada, Telia (Sweden, ;ritish Telecom (7G, 8rench Telecom (8rance, and

$mbratel (;ra-il. The international trials will have a single warehouse located

near 4ontreal, anada, and operated by Teleglobe, and 1> kiosks located at

participating signatory sites in Corth merica, $urope, and frica. The trialswill last for three months.

ommercial releases of I0S will add functionality to provide the following#

ontent"provider"based push of +eb content into the warehouse +eb

cache. I0S modules within the warehouse as well as the content provider

will be developed to implement this feature.



26/31


transport and pro%y for real"time streaming traffic within I0S.

Security and BoS for all traffic flowing through I0S.

;illing components for warehouses and kiosks.

Information mined from warehouse and kiosk databases to content

providers and service providers.

The first commercial version of I0S is scheduled for release in 4ay

1999. Two subse'uent commercial releases are planned in 1999.

Dept. of IT MESCE, Kuttippuram=


27/31


:. CONCLUSION

new generation of Internet access built around geosynchronous

satellites can provide immediate relief. They can improve service to bandwidth"

starved regions of the globe where terrestrial networks are insufficient and

supplement terrestrial networks elsewhere. This new generation of satellite

system manages a set of satellite links using intelligent controls at the link

endpoints. 4echanisms controlled include caching, dynamic construction of

push channels, use of multicast. and scheduling of satellite bandwidth.

Dept. of IT MESCE, Kuttippuram?


28/31


REERENCES

1. 6haleb bdulla, $dward . 8o%, 4arc brams, Shared 7ser ;ehavior

on the +orld +ide +eb, WebNet97, Toronto, 3ctober 199?,

http#JJwww.cs.vt.eduJNnrgJdocsJ9?webnetJ

. hen et. al, +ormhole aching with HTT! !ush 4ethod for a Satellite"

;ased 6lobal 4ulticast Eeplication System, submitted to the

International ache onference, San 0iego, une 1999.

/. !eter 0an-ig, Cetache rchitecture and 0eployment, 8ebruary ,

199?.

5. ames 6wert-man and 4argo Selt-er, +orld"+ide +eb ache

onsistency,http#JJwww.eecs.harvard.eduJvinoJuseni%.19=J

Dept. of IT MESCE, Kuttippuram:
http://www.cs.vt.edu/~nrg/docs/97webnet/http://www.eecs.harvard.edu/vino/usenix.196/http://www.eecs.harvard.edu/vino/usenix.196/http://www.cs.vt.edu/~nrg/docs/97webnet/http://www.eecs.harvard.edu/vino/usenix.196/


29/31


A;STRACT

Two scaling problems face the Internet today. 8irst, it will be years

before terrestrial networks are able to provide ade'uate bandwidth uniformly

around the world, given the e%plosive growth in Internet bandwidth demand and

the amount of the world that is still unwired. Second, the traffic distribution is

not uniform worldwide# lients in all countries of the world access content that

today is chiefly produced in a few regions of the world (e.g., Corth merica.

new generation of Internet access built around geosynchronous satellites can

provide immediate relief. The satellite system can improve service to

bandwidth"starved regions of the globe where terrestrial networks are

insufficient and supplement terrestrial networks elsewhere. This new generation

of satellite system manages a set of satellite links using intelligent controls at

the link endpoints. The intelligence uses feedback obtained from monitoring

end"user behavior to adapt the use of resources. 4echanisms controlled includecaching, dynamic construction of push channels, use of multicast, and

scheduling of satellite bandwidth. This paper discusses the key issues of using

intelligence to control satellite links, and then presents as a case study the

architecture of a specific system# the Internet 0elivery System, which uses

ICT$2ST)s satellite fleet to create Internet connections that act as wormholes

between points on the globe.



30/31


CONTENTS

1. ICTE307TI3C

. ISS7$S IC S$3C0"6$C$ETI3C ST$22IT$ 2ICG 3CTE32

/. S$ ST70K# ICT$EC$T 0$2IF$EK SKST$4

5. I0S EHIT$T7E$
http://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Introductionhttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#BM2_Issueshttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#BM3_Casehttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Roadmaphttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Referenceshttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Introductionhttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#BM2_Issueshttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#BM3_Casehttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#Roadmaphttp://www.isoc.org/inet99/proceedings/4q/4q_2.htm#References


31/31


ACKNO+LEDGE4ENT

I e%press my sincere gratitude to Dr. Agnisar#an Na#"oo'iri

adding interlligence to internet

Documents