Transcript 8 - 1

Chapter 8
Backbone Networks
8-1
Outline
• Components of Backbone networks
– Bridges, Routers, Gateways
• Backbone network architectures
• Backbone technologies
• Best practice backbone design
• Improving backbone performance
8-2
Backbone Networks
• High speed networks linking an
organization’s LANs
– Making information transfer possible between
departments
– Use high speed circuits to connect LANs
– Provide connections to other backbones,
MANs, and WANs
• Sometimes referred to as
– An enterprise network
– A campus-wide network
8-3
Backbone Network Components
• Network cable
– Functions in the same way as in LANs
– Optical fiber - more commonly chosen
(provides higher data rates)
• Hardware devices
– Computers or special purpose devices used
for interconnecting networks
• Bridges
• Routers
• Gateways
8-4
Backbone Network Devices
Device
Operates
at
Packets
Physical
Layer
Data Link
Layer
Filtered using
data link layer
addresses
Same
or
Different
Same
Same
Router
Network
Layer
Routed using
network layer
addresses
Same
or
Different
Same
or
Different
Same
Gateway
Network
Layer
Routed using
network layer
addresses
Same
or
Different
Same
or
Different
Same
or
Different
Bridge
Data Link Network
Layer
Layer
8-5
Bridges
• Data link layer devices
• Connect LANs with the same Data
Link and same Network layers
Allows different
types of cabling
Operate in a similar way to layer 2 switches (learning bridges)
8-6
Learning Bridges
• Operate in a similar way to layer 2 switches:
– Learn which computers are on each side of the bridge
• By reading the source addresses on incoming
frames and recording this information in forwarding
tables
• Data link layer devices
– Connecting similar type of networks
• But they can connect different types of cable
• Not popular anymore
– Losing market share to layer 2 switches as the latter
become cheaper and more powerful
8-7
Routers
• Operate at the network layer
• Connect LANS with different data
link layer, but the same network
layer protocol
Allows different
types of cabling
Perform more
processing than
bridges or layer 2
switches
8-8
Routers (Cont.)
• Operations
– Strip off the header and trailer of the incoming L2 frame
– Examine the destination address of the network layer
– Build a new frame around the packet
– Choose the “best” route for a packet (via routing tables)
– Send it out onto another network segment
• Compared to Bridges
– Perform more processing
• Process L3 messages (no changes made)
• Form new L2 messages for outgoing packets
– Processes only messages specifically addressed to it
8-9
Gateways
Also operate at network
layer (like routers)
Connect LANS
with different data
link layer and
different network
layer protocols
Some operate at
the application
layer as well
8 - 10
Other BB Network Devices
• Multiprotocol routers
– Can handle several different protocols (no translation)
• In and out protocols must be the same
• Brouters
– Combine bridge and router functions
• Examine L2 addresses of all messages
• Can also process directly addressed (L2) messages
• Layer-3 switches
– Similar to L2 switches, but switch messages based on
L3 addresses
– Can support many more simultaneous ports than
routers
8 - 11
Backbone Network Architectures
• Identifies the way backbone interconnects LANs
• Defines how it manages packets moving through BB
• Fundamental architectures
– Bridged Backbones
– Routed Backbones
– Collapsed Backbones
• Rack-based
• Chassis-based
– Virtual LANs
• Single-switch VLAN
• Multiswitch VLAN
8 - 12
Backbone Architecture Layers
• Access Layer (not part of BB)
– Closest to the users;
• Backbone Design Layers
– Distribution Layer
• Connects the LANs together (often in one building
– Core Layer (for large campus/enterprise networks)
• Connects different BNs together (building to building)
• <<<<<<<< Figure 8.5 goes here
8 - 13
Bridged Backbone
bus topology
Entire network is just one subnet
8 - 14
Bridged Backbones
• Move packets between networks based on their
data link layer addresses
• Cheaper (since bridges are cheaper than routers)
and easier to install (configure)
– Just one subnet to worry
– Change in one part may effect the whole network
• Performs well for small networks
– For large networks broadcast messages (e.g., address
request, printer shutting down) can lower performance
• Formerly common in the distribution layer
– Declining due to performance problems
8 - 15
Routed Backbone
Example of a routed BB at the Distribution layer
Usually a bus topology
Each LAN is a separate subnet
8 - 16
Routed Backbones
• Move packets using network layer addresses
• Commonly used at the core layer
– Connecting LANs in different buildings in the campus
– Can be used at the distribution layer as well
• LANs can use different data link layer protocols
• Main advantage: LAN segmentation
– Each message stays in one LAN; unless addressed
outside the LAN
– Easier to manage
• Main disadvantages
– Tend to impose time delays compared to bridging
– Require more management than bridges & switches
8 - 17
Collapsed Backbone
Most common type BB mainly used in
distribution layer
A connection to the switch is a
separate point-to-point circuit
Star topology
8 - 18
Collapsed Backbones
• Replaces the many routers or bridges of the
previous designs
– Backbone has more cables, but fewer devices
– No backbone cable used; switch is the backbone.
• Advantages:
– Improved performance (200-600% higher)
• Simultaneous access; :switched” operations
– A simpler more easily managed network – less devices
• Two minor disadvantages
– Use more and longer cables
– Reliability:
• If the central switch fails, the network goes down.
8 - 19
Rack-Based Collapsed Backbones
Figure 8-9 goes here
8 - 20
Rack-Based Collapsed Backbones
• Places all network equipment (hubs and switch)
in one room (rack room)
– Easy maintenance and upgrade
– Requires more cable (but cables are cheap)
• Main Distribution Facility (MDF) or Central
Distribution Facility
– Another name for the rack room
– Place where many cables come together
• Patch cables used to connect devices on the rack
• Easier to move computers among LANs
– Useful when a busy hub requires offloading
8 - 21
Main Distribution Facility (MDF)
• >>>> Figure 8.10 goes here
8 - 22
Chassis-Based Collapsed Backbones
• Use a “chassis” switch instead of a rack
– A collection of modules
• Number of hubs with different speeds
• L2 switches
• Example of a chassis switch with 710 Mbps capacity
– 5 10Base-T hubs, 2 10Base-T switches (8 ports each)
– 1 100Base-T switch (4 ports), 100Base-T router
–  ( 5 x 10) + (2 x 10 x 8) + (4 x 100) + 100 = 710 Mbps
• Flexible
– Enables users to plug modules directly into the switch
– Simple to add new modules
8 - 23
Virtual LANs (VLANs)
• A new type of LAN-BN architecture
– Made possible by high-speed intelligent switches
– Computers assigned to LAN segments by software
• Often faster and provide more flexible network
management
– Much easier to assign computers to different segments
• More complex and so far usually used for larger
networks
• Basic VLAN designs:
– Single switch VLANs
– Multi-switch VLANs
8 - 24
Single Switch VLAN Collapsed Backbone
acting as a large
physical switch
Switch
Computers assigned to
different LANs by software
8 - 25
Types of Single Switch VLANs
• Port-based VLANs (Layer 1 VLANs)
– Use physical layer port numbers on the front of the
VLAN switch to assign computers to VLAN segments
– Use a special software to tell the switch about the
computer - port number mapping
• MAC-based VLANs (Layer 2 VLANs)
– Use MAC addresses to form VLANs
– Use a special software to tell the switch about the
computer - MAC address mapping
• Simpler to manage
– Even if a computer is moved and connected to another
port, its MAC address determines which LAN it is on
8 - 26
Types of Single Switch VLANs
• IP-based VLANs (Layer 3 VLANs, protocol based
VLANs)
– Use IP addresses of the computers to form VLANs
– Similar to MAC based approach (use of IP instead of
MAC address)
• Application-based VLANs (Layer 4 VLANs, policybased VLANs)
– Use a combination of
• the type of application (Indicated by the port number
in TCP packet) and
• The IP address to form VLANs
– Complex process to make assignments
– Allow precise allocation of network capacity
8 - 27
Multi-switch VLAN-Collapsed Backbone
Switch
Switch
Switch
Switch
8 - 28
Multi-switch VLAN Operations
• Inter-switch protocols
– Must be able to identify the VLAN to which the packet
belongs
• Use IEEE 802.1q (an emerging standard)
– When a packet needs to go from one switch to another
• 16-byte VLAN tag inserted into the 802.3 packet by
the sending switch
– When the IEEE 802.1q packet reaches its destination
switch
• Its header (VLAN tag) stripped off and Ethernet
packet inside is sent to its destination computer
8 - 29
VLAN Operating Characteristics
• Advantages of VLANs
– Faster performance
• Precise management of traffic flow
• Ability to allocate resources to different type of
applications
– Traffic prioritization (via 802.1q VLAN tag)
• Include in the tag: a priority code based on 802.1p
• Can have QoS capability at MAC level
– Similar to RSVP and QoS capabilities at network and
transport layers
• Drawbacks
– Cost
– Management complexity
8 - 30
Backbone Technologies
• Gigabit Ethernet
• Fiber Distributed Data Interface (FDDI)
• Asynchronous Transfer Mode (ATM)
8 - 31
FDDI
• A set of standards designed in 80’s for MANs
(ANSI X3T9.5)
– Also used as BB and LAN technologies
• Limited future
– Gigabit Ethernet’s strong presence
• A ring network operating at 100 Mbps over fiber
cables
– Assumes a mix of 1,000 stations and 200 Km path
• With repeaters at every 2 Km
– Uses 2 counter rotating rings: primary and secondary
• Data on the primary; secondary used as backup
8 - 32
FDDI Topology
• >>>> Figure 8.15
Two types of
FDDI computers
secondary ring flows in opposite direction
8 - 33
Managing a Broken Ring in FDDI
If a ring is broken, the ring can still operate in a limited fashion
• >>>> Figure 8.16
8 - 34
FDDI Media Access Control
• Uses a controlled access token passing scheme
– Sending computer
• Wait for the token, when receive it
• Attach the packet to the token and transmit them
– Receiving computer
• See if there is a packet attached to the token
• If there is  process the packet
• If it needs to transmit a packet  follow the steps
above
• If no packet to send  simply transmit the token to
the next computer
• Very reliable and provide adequate response time
until it almost reaches saturation at 100 Mbps
8 - 35
ATM
• Originally designed for use in WAN
– Often used now in BNs
• Standardized; simple to connect BNs and WANs
• Also called cell relay
• Includes Layer 3, Layer 2 and Layer 1
technologies in the specifications
– Compatible with TCP/IP and Ethernet as if ATM was
Layer 2 technology
• A connection oriented technology
• ATM switches
– Provide point-to-point full duplex circuits at 155 Mbps
(622 Mbps for switch-to-switch)
8 - 36
ATM vs. Ethernet
• Packet format:
– Uses fixed-length packets (cells) of 53 bytes: 5-byte header,
48 byte data
– Designed to make switching faster (in hardware)
• Error Checking
– Error checking done for header only (not on data)
• If error detected, cell is discarded
• Addressing
– Uses a virtual channel(VC) between sender and receiver
• All cells use VC Identifier as addresses
• QoS (prioritized transmissions)
– Each VC assigned a specific class of service with a priority
8 - 37
Virtual Channels in ATM
• Identified by a two-part number
– Path number
– Circuit number within that path
• A physical port on a switch may have many paths
– A path may have many circuits
• A switch may have thousands of VCs
– A VC table is used to map the connections which can be
established either:
• Permanently: Permanent Virtual Circuit (PVC)
• Temporarily: Switched Virtual Circuit (SVC)
– Deleted when the connection is not needed
8 - 38
Addressing and Forwarding in ATM
When a cell arrives, switch checks the cell’s VC identifier at
the table and determines where to send it .
• >>>Figure 8.17 goes here
8 - 39
Approaches of Using ATM in Backbone
• LAN Emulation (LANE)
– Breaking LAN frame into 48-byte long blocks and
transmit them in an ATM cell
– Called encapsulation and done by edge switches
– Reassembling done at the destination edge switch and
LAN frame is sent to the LAN
– Requires translating of MAC addresses to VC Identifiers
(assuming VCs are setup already)
– Performance suffers due to encapsulation and
connection management
• Multiprotocol over ATM (MPOA)- LANE extension
– Uses IP addresses in addition to MAC addresses
• If same subnet, use MAC address; otherwise use IP
• ATM backbone operating like a network of brouters
8 - 40
Best Practice Backbone Design
• Architectures
– Performance and cost  Collapsed backbone
• VLANs closer; but not mature enough
• Efficiency of data rates
– Data Link Protocol Efficiency
• FDDI with 99%: Overhead 29 bytes; up to
4500 byte data
• ATM with about 87%: Overhead: 5 bytes
over 53 byte cell
– MAC Efficiency
8 - 41
FDDI MAC Efficiency
• Uses token passing controlled access
– Imposes more fixed-cost delays initially in low traffic
– Increases response times only slowly up to 90-95%
nominal capacity
– Total effective data rate = 89 Mbps
• 99% efficiency x 90% capacity x 100 Mbps
– >>>> Fig 8.19 goes here
8 - 42
ATM MAC Efficiency
• Uses full duplex transmission
– Efficiency ~ 100% of capacity
– Effective data rate = 135 Mbps each direction
simultaneously
• 87% efficiency x 100% capacity x 155 Mbps
• Total for both directions: 270 Mbps
– An ATM network with 622 Mbps circuits
• Provides 540 Mbps capacity each direction
•  1080 Mbps total
8 - 43
Conversion between Protocols
• Both requires conversion from/to Ethernet frames
• FDDI uses translation
– Remove Ethernet frame; replace it with FDDI frame
– Decreases efficiency 10-20%
– Actual total effective rate of FDDI  70 Mbps
• ATM uses encapsulation
– Segment and surround Ethernet frames with ATM cell
headers  Generally faster
– MAC Addresses must be translated to VC Identifiers and
VC management  30-40% decreased efficiency
– Actual total effective rate of ATM  80 Mbps each
direction (160 Mbps total)
8 - 44
Effective Data Rates of BB Technologies
• >>>Fig 8-20 goes here
8 - 45
Recommendations for BB Design
• Best architecture
– Collapsed backbone or VLAN
• Best technology
– Gigabit Ethernet
• Ideal design
– A mixture of layer-2 and layer-3 Ethernet switches
– Access Layer
• 10/100Base-T Later 2 switches with cat5e or cat6
– Distribution Layer
• 100base-T or 1000BaseT/F Layer 3 switches
– Core Layer
• Layer 3 switches running 10GbE or 40GBe
8 - 46
Best Practice BB Design
• >>>>>Fig 8-21 goes here
8 - 47
Improving Backbone Performance
• Improve computer and device performance
– Upgrade them to faster devices
– Use faster routing protocols
• Static routing is faster for small networks
– Use gigabit Ethernet as BB (eliminate translations)
– Increase memory in devices
• Improve circuit capacity
– Upgrade to a faster circuit; Add additional circuits
– Replace shared circuit BB with a switched BB
• Reduce network demand
– Restrict applications that use a lot of network capacity
– Reduce broadcast messages (placing filters at switches)
8 - 48
Implications for Management
• Increased traffic at backbone due to faster
technologies
– May requires that BN be replaced
–  Design BN to be easily upgradeable
• FDDI and ATM becoming as legacy technologies
– Vendors stopping the production of these
–  Begin to invest more funds to replace these
• Ethernet moving into Backbone extensively
– One standard technology used for both LANs and BN
–  Cheaper equipment; Easier management
8 - 49