Transcript Proposed Differentiated Services on the Internet

Proposed Differentiated
Services on the Internet
By
Dr. Junaid A. Zubairi
Department of Math and Computer Science
SUNY at Fredonia
Overview of Presentation








The Evolution of Internet
The Types of Traffic on the Internet
Performance Issues in Packet Switching
The World Wide Web
Changing Traffic on the Internet
The Birth and Death(?) of ATM!!
IETF’s DiffServ Model
MPLS
The Evolution of Internet
 DARPA (Defense Advanced Projects
Research Agency) started the ARPAnet
in 1969 between four nodes
 The Internet was primarily used for
connecting academic campuses together
until the mid 80’s
 Academicians used the Internet for
sharing ideas and research results via
email and ftp
The Evolution of Internet
 Internet was primarily used for email, usenet,
file transfer and remote usage of machines
 With email, users were able to keep in touch
 With usenet, users were able to discuss topics
of interest in focussed newsgroups
 With file transfer, users could download and
upload articles, programs and images
 With telnet, a user could login to his or her
UNIX account from anywhere
Types of Traffic on the Internet
 Email, usenet, ftp and telnet were
applications that generated almost similar
type of traffic stream on the Internet
 This traffic required “reliability”.
 The protocols were expected to deliver all the
data no matter how long it took
 With high error rate, the elapsed time simply
increased but the transmissions were
completed
Performance Issues in Packet
Switching
 For transmission on the Internet, the TCP/IP
suite of protocols breaks the data into
datagrams or packets and routes each packet
through an independently selected path
 Packets may arrive at the destination out of
sequence but due to buffering and reordering, the actual data can be recovered
easily
Performance Issues in Packet
Switching
 Path selection criterion is usually shortest
path first
 If the shortest path is congested or
unreliable, the router can choose another
path
 All routers decide independently and it is a
distributed environment
 Traditional TCP/IP based traffic is bursty
and it can increase or decrease abruptly
Performance Issues in Packet
Switching
 Given this scenario, a router may find itself
overwhelmed with a lot more packets than it
can handle
 Routers have limited buffering space in
which a queue of packets is managed (Refer
to Figure on next slide )
 Usually routers would use simple FIFO
scheme to select the next packet to be
transmitted
Adapted for academic use from
OPNET Modeler Tutorial
Performance Issues in Packet
Switching
 If the queue is full, the newly arrived
packets must be dropped (or discarded)
 Thus increase in traffic may increase timeouts, retransmissions and decrease in
efficiency
 To sum it up, the traditional TCP/IP network
based on packet switching is a “best effort”
network
Performance Issues in Packet
Switching
 The packet switching network makes its
best effort to deliver the data however it
makes no guarantees or promises to the user
about the network performance
 What is meant by network performance?
 Network performance is “Throughput”
 Throughput could be defined as the
aggregate rate of transmission offered by
the network to the user
The World Wide Web
 The introduction of hypertext marking
language (HTML) in early 1990 has
revolutionized the Internet
 HTML and its associated protocol HTTP
have transformed the Internet into a useroriented information repository
 HTML has also made it very easy to
“publish” information online even for users
with very little computer expertise
The World Wide Web
 The open ended HTTP has resulted in
supporting the linking of various types of
data into the web published documents
 HTTP makes it possible for web sites to
offer binary files, images, and multimedia
documents to the users with the click of a
button
 HTTP has also resulted in making the Internet
very popular. Internet continues to expand in
number of websites and the number of users
The World Wide Web
 Web deployment is flexible and easy
 Due to the web technologies, the Internet
has been put to use in almost all areas of
human knowledge
 For example, water distribution monitoring,
real-time traffic maps of big cities, free long
distance calling, distance learning with
lecture videos, buying and selling shares,
online shopping etc., the list appears endless
The Changing Traffic on the
Internet
 Due to the web enabled applications on the
Internet, there has been a tremendous
change in the types of traffic
 Now we have to deal with a significant
amount of traffic that is time-sensitive
 For example, consider the case of an audio
based application that needs to transmit the
data across the Internet
The Changing Traffic on the
Internet
Adapted for Academic Use from
"Computer Networks: A Systems
The Changing Traffic on the
Internet
Adapted for Academic Use from
"Computer Networks: A Systems
The Birth and Death(?) of ATM!!
 In this scenario, ATM offered a great
promise to the users
 ATM (Asynchronous Transfer Mode) is a
cell-switching technology that was targeted
to become the B-ISDN (Broad ISDN)
network of the future
 ATM was developed with the right targets
and it offered the much awaited
performance assurance
The Birth and Death(?) of ATM!!
 ATM is a connection-oriented technology
that offers various categories of services
(performance promises) to the users
 Service categories include
•
•
•
•
•
CBR (Constant Bit Rate such as telephony)
RT-VBR (real-time video such as videoconference)
NRT-VBR (Non real-time video e.g. stored video)
ABR (Available BR such as web browsing)
UBR (Unspecified BR such as ftp)
The Birth and Death(?) of ATM!!
 If a user requests a certain service, ATM
uses CAC (Connection Admission Control)
to determine if granting this request would
not jeopardize existing contracts
 User and network agree on certain QoS
parameters such as PCR (Peak Cell Rate),
SCR (Sustained Cell Rate) and CDV (Cell
Delay Variation) etc.
The Birth and Death(?) of ATM!!
 In order to meet the QoS contract
obligations, ATM network enforces traffic
shaping and policing
 Shaping involves techniques such as “Leaky
Bucket Algorithm” to regulate bursty traffic
 Policing means marking CLP (Cell Loss
Priority) on the offending cells that violate
the maximum rates agreed
Leaky Bucket Algorithm
Adapted for academic use from
"Computer Networks" Tanenbaum
The Birth and Death(?) of ATM!!
 However, the evolution of shared Ethernet
into switched Ethernet at 10Mbps and
development of Fast Ethernet at 100Mbps
stalled the ATM’s march to the desktop
 ATM was pushed back to the backbones of
campus networks
 Gigabit Ethernet in the backbone appears to
be the last nail in the coffin for ATM
The Birth and Death(?) of ATM!!
 ATM failed because of several factors




ATM is too complex (From packets to cells to
SONET frames, using AAL’s, emulating LAN)
ATM is expensive
Ethernet has evolved into much faster 100Mbps
and 1000Mbps services
All popular and established network
applications are packet based
IETF’s Models
 It was felt that instead of focussing on
coping with congestion, Internet should be
run in a way that there is no congestion
 Applications should be able to reserve
network resources at a given QoS
 IETF has been working on developing new
models and protocols for the Internet and
private networks
IETF’s Models
 IntServ and RSVP provide quantitative
guarantees to each flow. RSVP requires all
intermediate routers to keep track of each
and every flow through “soft state”.
 RSVP flows involve signaling and soft state
overhead and RSVP does not scale well to
the Internet. It may be successfully
deployed in a campus network
IETF’s DiffServ Model
 IETF is developing a model to provide
differing levels of service to different
applications without the overhead of
signaling and state maintenance
 The DiffServ model uses the TOS field in
IPv4 header to affix labels on packets
belonging to different service levels
 DiffServ has the potential to offer QoS on
the Internet, at last!!
IETF’s DiffServ Model
 Consider a gas station, you can buy regular,
super or premium gas from the same pump
 DiffServ offers various service levels to the
customer from the same network with SLA
 DiffServ adopts techniques used in ATM for
traffic management, in a simplified way
 DiffServ treats the network and the customer in
the way the ATM does.i.e. Customer gets a link at
the specified level of service to the network.
IETF’s DiffServ Model
 DiffServ levels of service are implemented
in a DiffServ domain
 The customer connects to the “edge router”
at the edge of the DiffServ domain
 The edge router performs traffic
classification (using DS codepoint marked
by customer in TOS to separate the packets)
 It then measures submitted traffic for
conformance to the agreed profile
IETF’s DiffServ Model
 The edge router then changes the DS code
byte of offending packets
 It may also do traffic shaping by delaying
the packets as necessary and dropping the
offending packets
 Refer to the diagram in the next slide to see
the edge router function
Adapted for Academic Use from
"Data and Computer
IETF’s DiffServ Model
 IETF has defined two DS services that are
visible as PHB (per-hop-behavior) of an
intermediate router for the marked packet
 EF (Expedited Forwarding)

EF is the premium service offered. It can
appear as a virtual leased line for the customer.
It offers low loss/latency and assured
bandwidth
 http://www.ietf.org/rfc/rfc2598.txt
IETF’s DiffServ Model
 AF (Assured Forwarding)

The AF PHB group provides delivery of IP
packets in four independently forwarded AF
classes. Within each AF class, an IP packet can
be assigned one of three different levels of drop
precedence. A DS node does not reorder IP
packets of the same microflow if they belong to
the same AF class.
 http://www.ietf.org/rfc/rfc2597.txt
MPLS
 MPLS (Multi Protocol Label Switching) is
the most promising protocol for the Internet
 An MPLS domain has an ingress node that
nails down paths through the maze of core
routers for every requesting flow until the
exit door (egress node)
 Thus every router does not have to decide
about the path of each packet
MPLS
 Intermediate routers use a “shim header” or
a layer 2.5 header to decide about the next
hop of a packet
 This shim header is inserted between the
frame header and packet header
 It is used by the router to consult a table that
tells what path is good for this packet
 This shim header is the “Label” and the
whole thing is called “Label Switching”
MPLS
 Instead of routing, now the routers do label
switching
 Since the path is pre-determined, routers
can speed up the processing of packets
 Also, the management can decide LSP’s
(label switched paths) based on load
distribution and other administrative goals
 Thus the connectionless network changes
into a connection oriented network
Summary
 Thus we can see that the Internet is
changing in a major way
 MPLS and Diffserv are being combined to
provide EF paths to certain flows such as IP
telephony, AF paths to multimedia
streaming and DF paths to ftp, email etc
 In future, Internet may be able to provide
the QoS that is only enjoyed by telephone
and Radio/TV broadcasting