Overview of IP routing

gascitytankNetworking and Communications

Oct 28, 2013 (4 years and 16 days ago)

94 views

Overview of IP routing
Tuomo Karhapää
Otaverkko Oy
tuomo.karhapaa@otaverkko.fi
© Tuomo Karhapää 2
Agenda
n Why routing protocols are needed?
n Routing in the Internet
n Designing large-scale IP
internetworks
n Introduction to routing protocols
n Case: MediaPoli
n Conclusions
© Tuomo Karhapää 3
Why routing protocols
are needed?
n IP network’s primary function is
deliver packets from source to
destination
n routing function is needed to know
what is the best route from source
to destination
n if network topology changes
routing function sets up a new
route from source to destination
© Tuomo Karhapää 4
Static vs. dynamic
routing
n static routes can be set by network
administrators
u “the hard way”
u mainly used as a default route
n dynamic routing
u route discovered and selected by
dynamic routing protocol
© Tuomo Karhapää 5
Address allocations
n Internet Assigned Numbers
Authority (IANA) allocates IP
addresses and AS-numbers
n ISPs get their IP addresses from
upstream provider or from regional
registry
u APNIC (Asian-Pacific Network
Information Center)
u ARIN (American Registry for
Internet Numbers)
u RIPE NCC (Réseaux IP
Européens)
© Tuomo Karhapää 6
Routing in the Internet
n Internet is set of autonomous
systems (AS)
n autonomous system is collection of
routers under same administration
n routing inside of AS is handled by
IGP (Interior Gateway Protocol)
n routing between ASes is done by
EGP (Exterior Gateway Protocol)
© Tuomo Karhapää 7
Routing in the Internet
R1
R8
R7
R2
R4
R3
R6
R5
AS1
Core
AS4AS3
AS2
© Tuomo Karhapää 8
Designing large-scale IP
internetworks
n design issues
u network topology
u addressing and route
summarization
u route selection
u convergence
u network scalability
u security
© Tuomo Karhapää 9
Network Topology
n physical topology
u set of routers
u networks which connect them
n logical topology
u different routing protocols establish
logical topology in different way
u flat topology vs. hierarcical
topology
© Tuomo Karhapää 10
Hierarchical network
Area 1 Area 3Area 2
Backbone
© Tuomo Karhapää 11
Addressing and route
summarization
n summarization reduces routing
information for each router
n several routes as a single
advertisement
n reduces the load on the router
n important when network size
increases
© Tuomo Karhapää 12
Route selection
R1
R4
R3
R5
R2
Source Dest
4
2
3
6
5
42
3
5
5
8
7
© Tuomo Karhapää 13
Metric calculation
n trivial if single path to destination
n computed by assigning a
characteristics to each physical link
(e.g.)
u link bandwidth
u link reliability
u propagation delay
n some protocols use multiple paths
with equal cost for load balancing
© Tuomo Karhapää 14
Convergence
n when the network topology
changes (e.g. link down), routers
need to
u detect the change
u select a new route
u inform other routers
n route selection is protocol-
dependent
© Tuomo Karhapää 15
Network scalability
n when network size increases,
some resources are critical
u memory
u CPU
u bandwidth
© Tuomo Karhapää 16
Security
n authentication prevents
unauthorized routers or hosts to
participate to routing process
n filters
u routes to be advertised
u routes not to advertised
© Tuomo Karhapää 17
Introduction to routing
protocols
n two categories
u distance vector protocol (e.g. RIP)
F broadcasts complete routing table
periodically and when network
topology changes
u link-state procol (e.g. OSPF)
F send table updates only when
change occurs
© Tuomo Karhapää 18
Distance vector
protocols
n each router calculates its routing
table
n shortest distance to networks or
routers are stored to routing table
n routers sends router table every 30
seconds to neighbors
n cannot scale
n cannot resolve loops quickly
© Tuomo Karhapää 19
Link-state protocol
n each router knows network
topology
n if topology changes, it is updated
by flooding the change to all
routers
n each router re-computes routing
table in parallel using link-state
database
© Tuomo Karhapää 20
RIP
n RIP = Routing Information Protocol
n distance vector protocol, IGP
n traditional routing protocol
n first version was shipped with BSD
distribution as a routed (1982)
n uses hop count as its metric
n the best route is which has lowest
metric
© Tuomo Karhapää 21
RIP
n max. hop count is 15 (16 is
unreachable)
n max. hop count is used to prevent
routing loops
n regular route table update causes
unnecessary resource use
n RIP is not suitable for low
bandwidth networks (e.g. frame
relay) or for large networks
© Tuomo Karhapää 22
RIPv2
n enhanced version of RIP
n distance vector protocol, IGP
n almost same as RIP
n supports variable-length subnet
masks and very simple
authentication
© Tuomo Karhapää 23
IGRP
n IGRP = Interior Gateway Routing
Protocol
n proprietary solution by Cisco
Systems in the mid-1980’s
n distance vector protocol
n it was designed to substitute RIP
© Tuomo Karhapää 24
EIGRP
n enhanced IGRP
n advanced distance vector protocol
n like link-state protocols EIGRP
sends updates only when changes
occurs
n supports variable-length subnet
masks
n route summarization at any bit
boundary
© Tuomo Karhapää 25
OSPF
n OSPF = Open Shortest Path First
n developed by IETF
n link-state protocol
n supports variable-length subnet
masks
n as IGP, OSPF is used inside a
single autonomous system (AS)
© Tuomo Karhapää 26
OSPF
n administration of large IP network
is simplified by dividing network to
areas
n mandatory area is the backbone
(area 0)
n backbone area is the transit
domain between other areas
© Tuomo Karhapää 27
OSPF
Area 1
Area 3
Area 2
Backbone
R4
R2
R5
R3
R1
R6
© Tuomo Karhapää 28
OSPF
n area border router (ABR) has
typically interface to local area and
to backbone area
n ABR summarizes information from
the local area to the backbone and
information is propagated to other
areas
© Tuomo Karhapää 29
OSPF
n when OSPF starts, it elects two
special routers
u designated router (DR)
u backup designated router (BDR)
n these routers has following
responsibilities
u determine which routers are
connected to network
u synchronize all routers’ link state
databases
© Tuomo Karhapää 30
OSPF
n some things are important when
designing OSPF network
u stability and redundancy of
backbone
u definition of area boundaries
u address assigment
u the number of routers per area
u selection of designated router
© Tuomo Karhapää 31
BGP4
n BGP = Border Gateway Protocol
n path vector protocol
n exchanges network reachability
information between BGP systems
n network reachability information
includes AS-path topology
n classless routing protocol
n supports route aggregation and
supernetting
© Tuomo Karhapää 32
BGP4
n learns multiple paths via internal
and external BGP peers
n selects the best path
n loop detection (important)
n policy based on AS path,
community or the network
u rejects/accepts selected routes
n no load balancing
© Tuomo Karhapää 33
BGP4: multi-homed
R2
R1
R3
R5
R4
AS200
AS100
AS300
© Tuomo Karhapää 34
Conclusion about
routing techniques
n recommendation is to
u use OSPF inside AS (not RIP)
u use BGP4 between autonomous
systems
n OSPF and BGP4 are commonly
used and it is possible to use them
in multivendor environment
n [E]IGRP is proprietary solution
© Tuomo Karhapää 35
Case:MediaPoli
n R&D program
n the prototype of the information
society of the next generation
n testbed for new network
technologies
n simulation environment for digital
media
n test and pilot environment for new
services and applications
n feasibility studies
© Tuomo Karhapää 36
Case: MediaPoli
n research units: e.g. HUT, VTT,
GSF, CSC
n students
n companies in Innopoli, in Science
and Technology Park and in
Spektri Business Park
n vendors
n content providers
n other companies
© Tuomo Karhapää 37
Case: MediaPoli
n backbone: Gigabit Ethernet
n access
u 10 Mbps switched Ethernet
u 100 Mbps switched Ethernet
u 1 Gbps Ethernet
u 2..11 Mbps WLAN (IEEE 802.11)
u ATM
n new technologies in the future
u xDSL, something new
© Tuomo Karhapää 38
Case: MediaPoli
n the production, marketing, delivery
and use of digital services
n distance education
n electronic commerce
n data security
n QoS routing
n 3rd generation’s mobile networks
n network technologies
© Tuomo Karhapää 39
Case: MediaPoli
n 4500 workstations connected to
the network
n at the moment most interesting
area is mobile networks
n after service pilots
commercialization of services
n on-going R&D work for network
and services
© Tuomo Karhapää 40
Case: MediaPoli,
network topology
Innopoli
CSC
HUT/Main
building
VTT
Spektri
GSF
Arcada
Housing
area for
students
HUT/CS-
building
Commercial ISP
Funet
Helsinki Telephony
Company /RC
HUT/EE
© Tuomo Karhapää 41
Case: MediaPoli
n superblock (32 c-classes, 8192 ip-
addresses)
u minimum block to advertise to the
Internet
u MediaPoli has one superblock from
RIPE
u addresses also from commercial
ISP
u own AS number
© Tuomo Karhapää 42
Case: MediaPoli
n Funet connection for govermental
organisations
n commercial ISP for companies
n OSPF inside campus area as IGP
n IBGP inside AS for better control
over routes
n EBGP for connections to upstream
providers’ networks
© Tuomo Karhapää 43
Case: MediaPoli
n Why OSPF and BGP4
u OSPF supports variable subnet
length (CIDR)
u OSPF is more efficient than RIPv2
u BGP4 provides loop-free routing
between autonomous systems
u BGP4 offers route maps for
filtering routes to advertise
© Tuomo Karhapää 44
Conclusions
n multi-homed network is not easy to
implement
n need to plan routing policies
carefully
n For more information
u MediaPoli: www.mediapoli.com
u Otaverkko: www.otaverkko.fi