P1: IML/FFX P2: IML/FFX QC: IML/FFX T1: IML
Web ̇QOS WL040/Bidgoli-Vol III-Ch-58 July 16, 2003 9:36 Char Count= 0
722 WEBQUALITY OFSERVICEconsistency, content quality, and access timeliness, ver-
sus resource constraints in the network such that some
notion of global utility is optimized. Such algorithms
will undoubtedly receive much attention in the near
future.
Although research in the Web QoS arena has been very
active, the state of deployment is lagging due to several
considerations primarily relating to the difficulty in de-
ploying integrated solutions that combine network QoS,
end-system QoS, and QoS-support in the operating sys-
tem. Fortunately, the networking community has made
dramatic strides towards making network support for
QoS a reality. This progress further increases the impor-
tance of solving the fundamental challenges in building
predictable architectures from the end system’s perfor-
mance perspective such that a next generation of Web
QoS systems finally becomes a reality. A Particularly good
reference for further readings on Web architecture,
performance, and QoS issues is the recent book by
Krishnamurthy and Rexford (2001).GLOSSARY
Authentication The act of verifying client identity.
Backbone provider A party that owns the communi-
cation fabric of an Internet backbone. Examples of
backbone providers include AT&T, Sprint, MCI, and
UUNET.
Cache server A network server that acts as a cache of
popular Web content and is able to serve it on behalf
of the original servers.
Content distribution network A network of server
platforms whose sole purpose is efficient content
dissemination around the Internet backbone.
Cookies Small text files that servers put on the client’s
hard drive to save client and session information
needed for future access.
Data pull model A data communication model in which
the client explicitly asks the server for data each time
the data are needed.
Data push model A data communication model in
which servers unilaterally push data to the client with-
out being asked. The model is a good optimization
when future client requests can be accurately pre-
dicted.
Demultiplexing Separating an incoming packet flow
into multiple segregated flows. For example, demul-
tiplexing must occur upon the arrival of a packet at a
server, in order to queue the packet for the right recip-
ient.
Differentiated services A framework for classifying
network traffic and defining different policies for han-
dling each traffic class, such that some classes receive
better service than others.
EDF Earliest-deadline-first scheduling policy. As the
name suggests, it schedules the task with the earliest
deadline first.
HTML Hypertext markup language, a language for
defining the content and appearance of Web pages.
HTTP Hypertext transfer protocol. It is the protocol
used for Web access in the current Internet. Currently,
it has two popular versions, HTTP 1.0 and HTTP 1.1.IP Internet protocol. It is the glue that connects the com-
puter subnetworks of which the Internet is composed
and is responsible for packet addressing and routing
between Internet senders and receivers.
IP fragment Part of an IP-layer message after fragmen-
tation.
Kernel The core part of the operating system, typically
responsible for scheduling and basic interprocess com-
munication.
Microkernel An operating system architecture where
most operating system functions are delegated to user-
level processes, keeping the kernel small.
QoS Quality- of service, a term used in quantifying dif-
ferent performance aspects of Web access such as time-
liness and throughput.
Packet A unit of data transfer across a network.
Persistent connections Communication abstraction
implemented by HTTP 1.1. It allows the same TCP
connection to be reused by multiple Web requests
to the same server. This is a main departure from
the traditional “one request per connection” model of
HTTP 1.0.
Proxy server A specialized server that performs an aux-
iliary Web content management function such as con-
tent replication, caching, or transcoding.
Semaphore An operating system construct used for
synchronization.
Sockets The main interprocess communication
abstraction, originally introduced in UNIX. A socket
represents a connection endpoint. The connection
is between two processes on the same or different
machines.
Threads The smallest schedulable entities in multi-
threaded operating systems.
TCP Transmission control protocol, the transport pro-
tocol used for reliable data communication on the
Internet.
Transcoding The process of converting content on the
fly from the server’s format to a format more suitable to
the client, or more appropriate for network load con-
ditions.
Web hosting The business of providing resources
(servers, disk space, etc.) to serving customers’ Web
pages. Typically, Web-hosting companies build large
server installations of hundreds of machines for serv-
ing the Web sites. These installations are called server
farms.CROSS REFERENCES
See Circuit, Message, and Packet Switching; HTML/
XHTML (HyperText Markup Language/Extensible Hyper-
Text Markup Language); Middleware; TCP/IP Suite; Web
Hosting.REFERENCES
Abdelzaher, T. F., & Lu, C. (2001). Schedulability anal-
ysis and utilization bounds for highly scalable real-
time services. Paper presented at the IEEE Real-
Time Technology and Applications Symposium, Taipei,
Taiwan.