Transcript ATM - La Salle University

ATM
Computer Networks and Internets,
Comer
CSIT 220 (Blum)
1
Once you have a path
• Scheme 1: once found, the path does not change,
and channels within that path are dedicated to the
particular communication exchange, data (can be)
continuous
– Circuit switching
• Scheme 2: once found, the path does not change,
but channels are shared, data broken into packet
– Connection-oriented packet switching
• Scheme 3: paths may change, channels are shared,
data broken into packets
– Connectionless packet switching
CSIT 220 (Blum)
2
ATM
• One example of Scheme 2 on the previous slide is
ATM.
• ATM: Asynchronous Transfer Mode
• “a … high-speed, full-duplex, connectionoriented, fixed-length cell-switching scheme that
is suitable for data as well as digitized voice and
video.” (Computer Dictionary)
• ATM grew out of the ISDN standards and is part
of B-ISDN (broadband-ISDN).
CSIT 220 (Blum)
3
Trying to be all things
• ATM had very ambitious goals. It was designed to
handle voice and video as well as data
– The voice/video put strong constraints on the speed and
steadiness of the transmission rate.
– The connection-oriented and small, fixed cell size
decisions are both motivated by the voice/video
constraints.
• It was designed to be both a LAN and a WAN
technology.
– The emergence of faster Ethernet technologies is
pushing ATM out of the LAN market but it remains a
viable WAN technology.
CSIT 220 (Blum)
4
Connectionless vs. Connection-Oriented
• A “connection” is not only setting up a physical
and/or logical path but also an agreement
between the sender and receiver to communicate.
– Connectionless means that one can begin to transmit
data without establishing a “connection”
• e.g. sending (snail) mail or IP (Internet Protocol)
– Connection-oriented requires a connection
• e.g. calling on the phone or TCP (Transmission Control
Protocol)
CSIT 220 (Blum)
5
Connection
• Connections used to
be made by physically
plugging wires into a
switchboard.
• Early computer
programming was
done in a similar
fashion.
CSIT 220 (Blum)
6
ENIAC
CSIT 220 (Blum)
7
Virtual circuit
• Analogous to the way programs moved from
hardwiring to RAM (and some ROM),
establishing communication channels also became
more software based.
• The software-based connection from source to
destination is known as a “virtual circuit.”
– There must be a physical connection. Actually there are
many possible physical connections and the software
selects the desired path.
CSIT 220 (Blum)
8
PVC and SVC
• Somewhat analogous to programming in which we
have ROM (for long-term memory) and RAM (for
short-term memory), in virtual circuits we have
permanent virtual circuits (PVC) and switched
virtual circuits (SVC).
– A PVC has to be “configured” (a somewhat slow
process) by the company providing the connection. It is
permanently available (nailed down) to the user.
Analogous to ROM.
– An SVC is set up more quickly and exists only during a
“session.” Analogous to RAM.
CSIT 220 (Blum)
9
Establishing an connection
• In the connection-oriented approach, transmission
is requested and accepted.
• Recall the DB9 connector (from homework), in
addition to the Receive Data, Transmit Data, there
were pins such as Carrier Detect, Request to Send,
Clear to Send, etc. These are control pins used in
“handshaking” which is establishing a connection.
– (This handshaking is built into in the connector making
this a hardware connection.)
CSIT 220 (Blum)
10
Establishing an connection II
• You may have noticed in another homework that
some of the ASCII characters are used in
“handshaking” as well.
– (software connection)
• Another place we have seen a hint of this
connection establishing is in ISDN which had
three channels, a D and two B
– The D channels request service over the B channels.
– The control is separated from the data and is said to be
“out of band.”
CSIT 220 (Blum)
11
Establishing an connection III
• A packet-switched, connection-oriented system
like ATM establishes an SVC, by having the
source send out control packets.
• As the control packet is routed to its destination, it
establishes the virtual circuit by entering
information into the tables of the ATM switches
(the “routers” of an ATM network).
• If the destination “accepts” the “request” it returns
an accept packet.
• The subsequent data packets are identified not by
a destination address but by a virtual channel
identifier (VCI).
CSIT 220 (Blum)
12
VPI/VCI
• In place of source and destination addresses,
ATM has VPI (Virtual Path Identifier) and
VCI (Virtual Channel Identifier).
• Recall that in broadband connections, the
line is broken down into many channels, the
VPI determines the route taken (from
ATM switch to ATM switch) and the VCI
determines the channels used within those
connections.
CSIT 220 (Blum)
13
Comer Fig. 14.2
CSIT 220 (Blum)
14
Forwarding
• The ATM switches have forwarding tables.
• Instead of looking through a table each
time, part of the entry in one ATM switch’s
table is the location for the same virtual
channel in the next ATM switch’s table.
• This was set up when the virtual circuit was
negotiated.
CSIT 220 (Blum)
15
Comer Fig. 14.3
CSIT 220 (Blum)
16
Label switching
• The forwarding table gives the (physical)
port of the ATM switch through which one
should leave and the location of the virtual
channel in the next forwarding table.
• In this approach, the data field that directs
the packet changes at each ATM switch, this
process is known as “label rewriting” or
“label switching.”
CSIT 220 (Blum)
17
UNI-NNI
• The VPI/VCI part changes nature somewhat
at the last stage when the cell is about to
leave to the network and arrive at a
computer.
– NNI (network to network interface) between
ATM switches and other ATM switches.
– UNI (user to network interface) between
ATM switches and computers.
CSIT 220 (Blum)
18
The connection-oriented decision
• Setting up a virtual circuit is not worthwhile if one
is sending only a few packets.
• It becomes worthwhile when one intends to send a
steady (or nearly steady) stream of data between a
given source and destination.
• In “live” or real-time audio/video transmission,
delay becomes intolerable.
• The choice in ATM cell size had similar
motivations.
CSIT 220 (Blum)
19
Fixed-Length Cell
• Previously we have seen variable payloads
– Ethernet up to 1500 bytes
– Frame Relay up to 4096 bytes
• ATM fixes the payload size to a relatively
small value of 48 bytes
– When packets are fixed in size, they are called
“cells”
CSIT 220 (Blum)
20
The Size Compromise
• 48 bytes was a compromise between Europe
who wanted 32 and US/Japan who wanted
64.
• The size is also a compromise between data
transmission where large packets are
preferred and (live) audio/video where
small packets are preferred.
CSIT 220 (Blum)
21
ATM cell (Fig. 14.1)
CSIT 220 (Blum)
22
Percentage
• An ATM cell has 53 bytes altogether
• The first 5 bytes comprise the header
– The VPI/VCI portion is the addressing
• The last 48 bytes comprise the payload
• The payload is 90.6% of the cell
48
48
=
 90.6%
48+5
53
• (Payload is the complement of overhead, which in
this case is 9.4%.)
CSIT 220 (Blum)
23
Compare to Ethernet
• Compare this to the largest Ethernet packet,
which has a 1500-byte payload and 26 bytes
of header/trailer (including the preamble)
1500
1500
=
1500+26
1526
CSIT 220 (Blum)
 98.3%
24
Cell Tax
• The percentage of the cell/packet that is not
“actual” data (but protocol data) is
significantly higher in ATM compared to
Ethernet or Frame relay.
• It has a higher “overhead.”
• This feature is sometimes called the “cell
tax.”
CSIT 220 (Blum)
25
Small but Fast
• Having a small, fixed size
– reduces the overhead involved in transmitting
cells, this reduces the “latency” — the time
delay associated with a switch
• Managing cells is simplified by their fixed size.
• Store and forward is faster if there’s less to store and
forward (at a given time).
• More done by hardware, less by software, speeds it
up.
– Leads to more predictable transmission rates.
CSIT 220 (Blum)
26
Hardware based
• The pre-specified bit rates are either
155.520 Mbps or 622.080 Mbps.
• Speeds on ATM networks can reach 10
Gbps.
• Along with Synchronous Optical Network
(SONET) and several other technologies,
ATM is a key component of broadband
ISDN (BISDN).
CSIT 220 (Blum)
27
jitter
• The small size is used to prevent delays in
the transmission of real-time audio/video.
• Video and audio require a fairly steady rate.
• If a video frame is missing or delayed, the
rate is uneven, and the image is said to
“jitter.”
• Long-distance communication involving
satellites often has a delay, making
communication strained.
CSIT 220 (Blum)
28
IMA
• Inverse Multiplexing over ATM takes a stream
of ATM cells and breaks it up, and transmits it
across multiple T1/E1 links, then reconstructs it in
the original ATM cell order at the destination.
• Alternative for those who cannot afford a T3 line.
• IMA is a User-to-Network Interface standard
approved by the ATM Forum in 1997.
CSIT 220 (Blum)
29
Video conferencing
• A “meeting” between two or more people at
different sites by using computer networks to
transmit audio and video data.
• Until the mid 90s, videoconferencing was too
expensive for most businesses, but that is
changing fast.
• “Many analysts believe that videoconferencing
will be one of the fastest-growing segments of the
computer industry in the latter half of the decade.”
CSIT 220 (Blum)
30
The quality of service is not strained
• QoS stands for Quality of Service, which is when
a network guarantees a certain throughput.
• That ATM supports a QoS level is a big advantage
that it has over Frame Relay and Fast Ethernet.
• ATM providers guarantee their customers a certain
amount of bandwidth and a minimal amount of
that (end-to-end) latency.
CSIT 220 (Blum)
31
Best effort
• Transmitting real-time video/audio dependably is
difficult in public networks using ordinary "best
effort" protocols.
• Using the Internet's Resource Reservation
Protocol (RSVP), packets passing through a
gateway host can be expedited based on policy
and reservation criteria arranged in advance.
• But no guarantees like ATM.
CSIT 220 (Blum)
32
ATM: Service Type
• Constant Bit Rate (CBR): a fixed rate, data is
sent steadily. This is analogous to a leased line.
• Variable Bit Rate (VBR): a specified throughput
capacity, but data sent somewhat unevenly. This is
a popular choice for videoconferencing (need
compression)
• Unspecified Bit Rate (UBR) no guaranteed
throughput levels. This is used for applications,
such as file transfer, that can tolerate delays.
• Available Bit Rate (ABR) guaranteed minimum
capacity but allows data to be bursted at higher
capacities when the network is free.
CSIT 220 (Blum)
33
Compression
• The variable bit rate is acceptable if the
audio/video signal is compressed.
• Text example: instead of representing all letters by
the same number of bits as in ASCII, one can
represent common letters like E as shorter strings
and less common letters like Q as longer strings so
that on average the number of bits that must be
sent is reduced
– Lossless: no information lost, just sent or stored more
efficiently
– Lossy: some information is lost, but it was considered
superfluous
CSIT 220 (Blum)
34
References
•
•
•
•
http://www.webopedia.com
http://www.whatis.com
Computer Dictionary, Mitchell Shnier
Understanding Data Communications &
Networks, Shay
CSIT 220 (Blum)
35