Routing Within An Routing Within An Autonomous SystemAutonomous System
(RIP, OSPF)(RIP, OSPF)
Chapter 15Chapter 15
Static vs Dynamic Interior RoutesStatic vs Dynamic Interior Routes
Interior RoutersInterior Routers Two routers within an autonomous systemTwo routers within an autonomous system Are considered to be interior to one anotherAre considered to be interior to one another
How do they learn about their own networks?How do they learn about their own networks? Small, slowly changing systemsSmall, slowly changing systems
Establish and modify routes by handEstablish and modify routes by hand
Update table when new network added or deletedUpdate table when new network added or deleted
Figure…routing is trivialFigure…routing is trivial
Figure 15.1Figure 15.1
Disadvantages of manual systemDisadvantages of manual system Cannot accommodate rapid growthCannot accommodate rapid growth Cannot accommodate rapid changeCannot accommodate rapid change
Need automated methodsNeed automated methods Respond to change more easilyRespond to change more easily Improve reliabilityImprove reliability Better response to failuresBetter response to failures
Figure 15.2Figure 15.2
Multiple physical pathsMultiple physical paths Usually pick one to be primaryUsually pick one to be primary Router(s) fail along primary, must changeRouter(s) fail along primary, must change Manually: time-consuming & error proneManually: time-consuming & error prone Even in small internets need automated systemEven in small internets need automated system
For automationFor automation Interior routers must communicateInterior routers must communicate Exchange routing informationExchange routing information Once data established, advertiseOnce data established, advertise
One interior router advertises to other autonomous One interior router advertises to other autonomous systems via Exterior Gateway Protocolsystems via Exterior Gateway Protocol
No single interior protocol has emergedNo single interior protocol has emerged Varied topologies and technologiesVaried topologies and technologies Tradeoffs between simplicity & functionalityTradeoffs between simplicity & functionality
Easy to install & configure; less functionalityEasy to install & configure; less functionality Multiple protocols have become popularMultiple protocols have become popular
Small ASSmall AS Choose a single one; use exclusively internallyChoose a single one; use exclusively internally
Larger ASLarger AS Often choose a small setOften choose a small set
Interior Gateway ProtocolInterior Gateway Protocol Used as generic descriptionUsed as generic description Refers to any algorithm used by interior routersRefers to any algorithm used by interior routers Routers may run BGP to advertise reachabilityRouters may run BGP to advertise reachability
Need an IGP to obtain information within the ASNeed an IGP to obtain information within the AS
Figure 15.3Figure 15.3
Routing Information Protocol (RIP)Routing Information Protocol (RIP)
One of most widely used IGPsOne of most widely used IGPs Also know as route-dAlso know as route-d Came from Univ of CA – BerkeleyCame from Univ of CA – Berkeley
Developed for machines on their LANsDeveloped for machines on their LANs Relies on physical network broadcast to make Relies on physical network broadcast to make
routing exchanges quicklyrouting exchanges quickly Not designed to use on large WANsNot designed to use on large WANs
Versions of RIP adapted for WANs are soldVersions of RIP adapted for WANs are sold
Popularity not only due to technical meritsPopularity not only due to technical meritsWas distributed with popular 4BSD UNIX systemsWas distributed with popular 4BSD UNIX systems
Lots of TCP/IP sites used without even considering Lots of TCP/IP sites used without even considering technical aspectstechnical aspects
Once installed it became the basis for local routingOnce installed it became the basis for local routing RIP was built and adopted without a standardRIP was built and adopted without a standard
Most implementations derived from Berkeley codeMost implementations derived from Berkeley code
Interoperability was limitedInteroperability was limited Many undocumented details and subtletiesMany undocumented details and subtleties
New versions led to more problemsNew versions led to more problems
Standard appeared in June 1988Standard appeared in June 1988
RIP OperationRIP Operation Straightforward distance-vector routingStraightforward distance-vector routing Partitions participants into two categoriesPartitions participants into two categories
Active: Active: Advertise their routesAdvertise their routes Only routersOnly routers
Passive:Passive: Do not advertiseDo not advertise Host must use passive modeHost must use passive mode
Active RIP routers broadcast every 30 secondsActive RIP routers broadcast every 30 secondsSends routing update messageSends routing update message
Takes information from current routing databaseTakes information from current routing database Update is a set of pairsUpdate is a set of pairs
(IP address, integer distance to that network)(IP address, integer distance to that network)
Hop count is used as the distance metricHop count is used as the distance metric
One hop: directly connectedOne hop: directly connected
Two hops: reachable through one other routerTwo hops: reachable through one other router
Hops = number of networks datagram will encounterHops = number of networks datagram will encounter
Hop count for shortest path not always optimalHop count for shortest path not always optimal 3 Ethernets faster than 2 satellites3 Ethernets faster than 2 satellites Some RIP implementations allow assignment of artificially Some RIP implementations allow assignment of artificially
high hop counts when advertising slow networkshigh hop counts when advertising slow networks
Active & passive routers listen to broadcastsActive & passive routers listen to broadcastsAll update tables according to DV algorithmAll update tables according to DV algorithm
May take some time for advertisements to propagateMay take some time for advertisements to propagate Routers use Routers use hysteresishysteresis to improve performance to improve performance
Does not replace a route with an equal cost routeDoes not replace a route with an equal cost route
Prevents oscillation among equal pathsPrevents oscillation among equal paths Timers are used on all routesTimers are used on all routes
Solves problem of routes through crashed routersSolves problem of routes through crashed routers
Start timer when install route in tableStart timer when install route in table
Restart whenever receive msg advertising that routeRestart whenever receive msg advertising that route
Route is invalid if 180 seconds pass without another Route is invalid if 180 seconds pass without another advertisementadvertisement
RIP must handle three types of errorsRIP must handle three types of errorsRouting loopsRouting loops
Algorithm does not explicitly detect routing loopsAlgorithm does not explicitly detect routing loops Must either assume participants can be trusted or take Must either assume participants can be trusted or take
precautions to preventprecautions to prevent
InstabilitiesInstabilities Must limit hop count to preventMust limit hop count to prevent Maximum possible distance value is 16Maximum possible distance value is 16 If legitimate hop count is higher, must divide the internet into If legitimate hop count is higher, must divide the internet into
sections or use an alternate protocolsections or use an alternate protocol
Slow convergenceSlow convergence Routing messages propagate slowly across the networkRouting messages propagate slowly across the network Can lead to inconsistenciesCan lead to inconsistencies
Not unique to RIP; fundamental problem in any DV Not unique to RIP; fundamental problem in any DV protocolprotocol
Hop count limit helps but does not eliminateHop count limit helps but does not eliminate
Figure 15.4Figure 15.4
Solving slow convergenceSolving slow convergence ““Good news travel quickly; bad news travel Good news travel quickly; bad news travel
slowly”slowly”Quick to install good routeQuick to install good routeUnreachable only after timeout; then learn and Unreachable only after timeout; then learn and propagate new routepropagate new route
Split horizon updateSplit horizon updateRouter does not propagate information over the Router does not propagate information over the interface from where the route arrivedinterface from where the route arrived
RR22 would not advertise route to network would not advertise route to network 11 to to RR11
If If RR11 loses connectivity, it must quit advertising loses connectivity, it must quit advertising
After a few rounds of routing updates, all routers After a few rounds of routing updates, all routers agree that the network is unreachableagree that the network is unreachableDoes not prevent all routing loopsDoes not prevent all routing loops
Hold downHold downRouter ignores information for a period of timeRouter ignores information for a period of time
Typical time is 60 secondsTypical time is 60 seconds
Done after receiving msg that network is unreachableDone after receiving msg that network is unreachable
Wait so all machines can get bad news; keeps from Wait so all machines can get bad news; keeps from mistakenly accepting an out-of-date messagemistakenly accepting an out-of-date message
All machines must have same idea of hold downAll machines must have same idea of hold down Otherwise, get routing loopsOtherwise, get routing loops
Disadvantages:Disadvantages: If routing loop occurs, will be preserved for the hold downIf routing loop occurs, will be preserved for the hold down Also preserves incorrect routes during the hold down timeAlso preserves incorrect routes during the hold down time
Even when alternatives existEven when alternatives exist
Poison reversePoison reverseWhen connection disappearsWhen connection disappears
Advertising router keeps entry for few update periodsAdvertising router keeps entry for few update periods Puts infinite cost in the broadcastsPuts infinite cost in the broadcasts
Combine with triggered updatesCombine with triggered updates Router sends immediate broadcast when get bad newsRouter sends immediate broadcast when get bad news Not wait for next broadcast timeNot wait for next broadcast time Minimizes time it is vulnerable to believing good newsMinimizes time it is vulnerable to believing good news
These techniques solve some problems; These techniques solve some problems; introduce othersintroduce others
Triggered updatesTriggered updates Suppose many routers share common networkSuppose many routers share common network Single broadcast changes all tables; triggering moreSingle broadcast changes all tables; triggering more Broadcast avalancheBroadcast avalanche
BroadcastsBroadcasts Take substantial bandwidth themselvesTake substantial bandwidth themselves
Loops prevent stopping loopsLoops prevent stopping loops Looping messages may prevent routing msgs to break loopsLooping messages may prevent routing msgs to break loops
Hold down in WANsHold down in WANs Period so long, higher level protocol timers may expirePeriod so long, higher level protocol timers may expire Breaks the connectionsBreaks the connections
RIP1 message formatRIP1 message format Messages are of two typesMessages are of two types
Routing information messagesRouting information messages Periodic broadcast of unsolicited response messagesPeriodic broadcast of unsolicited response messages
Messages to request informationMessages to request information Routers or hosts can ask for info with request commandRouters or hosts can ask for info with request command Routers reply using a response commandRouters reply using a response command
Both use same formatBoth use same format
CommandCommand MeaningMeaning
11 Request for partial or full routing informationRequest for partial or full routing information
22 Response (network-distance pairs from sender’s routing table)Response (network-distance pairs from sender’s routing table)
99 Update Request (used with demand circuits)Update Request (used with demand circuits)
1010 Update Response (used with demand circuits)Update Response (used with demand circuits)
1111 Update Acknowledge (used with demand circuits)Update Acknowledge (used with demand circuits)
Figure 15.5Figure 15.5
RIP2 Address ConventionsRIP2 Address Conventions Skip…..Skip…..
RIP route interpretation and aggregationRIP route interpretation and aggregation Version 1 contains no provision for subnet maskVersion 1 contains no provision for subnet mask
Originally designed for classful addressingOriginally designed for classful addressing
Extended to allow subnettingExtended to allow subnetting
Important restriction:Important restriction: Subnet routes can only go in updates sent across networks Subnet routes can only go in updates sent across networks
that are part of the subnetted prefixthat are part of the subnetted prefix Cannot use with variable-length subnet addresses or Cannot use with variable-length subnet addresses or
classlessclassless Due to not having explicit subnet mask informationDue to not having explicit subnet mask information
May have updates for networks in & out of prefixMay have updates for networks in & out of prefix Router must prepare different update messagesRouter must prepare different update messages
RIP2 extensionsRIP2 extensions Contains provisions for explicit subnet maskContains provisions for explicit subnet mask Also include explicit next-hop informationAlso include explicit next-hop information
Prevents routing loopsPrevents routing loops
Prevents slow conversionPrevents slow conversion
RIP2 message formatRIP2 message format Puts new info in unused octets of address fieldPuts new info in unused octets of address field
Router can use both versions simultaneouslyRouter can use both versions simultaneously
Version number in same octet; inspect before processVersion number in same octet; inspect before process Adds 16-bit ROUTE TAGAdds 16-bit ROUTE TAG
Identify the origin of the routeIdentify the origin of the route
Figure 15.6Figure 15.6
Transmitting RIP messagesTransmitting RIP messages Messages do not have explicit length fieldMessages do not have explicit length field
Nor any explicit count of entriesNor any explicit count of entries Rely on delivery mechanism to tell lengthRely on delivery mechanism to tell length
With TCP/IPWith TCP/IP Rely on UDP to tell receiver the message lengthRely on UDP to tell receiver the message length RIP operates on UDP port 520RIP operates on UDP port 520
Disadvantage of RIP hop countsDisadvantage of RIP hop counts RIP restricts routing to a hop-count metricRIP restricts routing to a hop-count metric RIP restricts size of any internet using itRIP restricts size of any internet using it
Has small hop count value for infinity (16)Has small hop count value for infinity (16)
Limits span to at most 15 routers between hostsLimits span to at most 15 routers between hosts
Is not a limit on total number or density of routersIs not a limit on total number or density of routers In any case, hop count is a crude measureIn any case, hop count is a crude measure
Not always get least delay or highest capacity routesNot always get least delay or highest capacity routes
Makes routes static; cannot change due to loadMakes routes static; cannot change due to load
The Hello ProtocolThe Hello Protocol
IGP that uses routing metric other than hopsIGP that uses routing metric other than hops Now obsoleteNow obsolete Historically, used in original NSFNET backbone Historically, used in original NSFNET backbone
“fuzzball” routers“fuzzball” routers Uses metric of delayUses metric of delay Provides two functionsProvides two functions
Synchronizes clocks among a set of machinesSynchronizes clocks among a set of machines
Allows each machine to compute shortest delay pathsAllows each machine to compute shortest delay paths
Messages carry timestamp as well as routing infoMessages carry timestamp as well as routing info
Each participating machine maintains tableEach participating machine maintains table
Contains best estimate of neighboring machine clocksContains best estimate of neighboring machine clocks
Transmit timestamp with each packetTransmit timestamp with each packet Receiver computes estimate of delay on the link by using the Receiver computes estimate of delay on the link by using the
timestamp and its estimate of the sender’s clocktimestamp and its estimate of the sender’s clock
Periodically poll neighbors to update clock estimatesPeriodically poll neighbors to update clock estimates Standard D-V approach for updateStandard D-V approach for update
Send table of destinations & estimated delaysSend table of destinations & estimated delays
Receiver’s update tables if cheaper route advertisedReceiver’s update tables if cheaper route advertised
Delay Metrics & OscillationDelay Metrics & OscillationIs delay a good routing metric?Is delay a good routing metric? Would seem soWould seem so Worked well in the early Internet backboneWorked well in the early Internet backbone InstabilityInstability is the reason most protocols do not is the reason most protocols do not
use delayuse delayAny protocol that changes routes quickly can become Any protocol that changes routes quickly can become unstableunstable
Hop counts fixed; delay is notHop counts fixed; delay is not Minor variations in delay measurements occurMinor variations in delay measurements occur
Hardware clock driftHardware clock drift
CPU load during measurementCPU load during measurement
Delays by link-level synchronizationDelays by link-level synchronization If react quickly to slight variations, get two-If react quickly to slight variations, get two-
stage oscillationstage oscillationSwitch back and forth between alternate pathsSwitch back and forth between alternate paths
Heuristics to help avoid oscillationHeuristics to help avoid oscillation Hold downHold down
Slows down changingSlows down changing Round off measurements or use thresholdRound off measurements or use threshold
Ignore differences less than the thresholdIgnore differences less than the threshold Use average measurementUse average measurement
Keep average of recent measurementsKeep average of recent measurements
Use Use K-out-of-NK-out-of-N rule rule KK of the most recent of the most recent NN measurements must be less than the measurements must be less than the
current delay before route can be changedcurrent delay before route can be changed
Can still have instabilityCan still have instability Due to comparing delays on paths with Due to comparing delays on paths with
different characteristicsdifferent characteristicsTraffic has dramatic effect on delayTraffic has dramatic effect on delay
As load increases, delay grows rapidlyAs load increases, delay grows rapidly Fall into Fall into positive feedback cyclepositive feedback cycle
Burst of traffic at one place increases delayBurst of traffic at one place increases delay
Protocol changes routeProtocol changes route
New traffic may cause another change in delayNew traffic may cause another change in delay Another route change occursAnother route change occurs
Must have mechanism to dampen oscillationMust have mechanism to dampen oscillation
Previous heuristics may not helpPrevious heuristics may not help They help in simple case for paths with same They help in simple case for paths with same
throughput characteristicsthroughput characteristics Not good when paths have different delay and Not good when paths have different delay and
throughput characteristicsthroughput characteristicsCompare serial line and satellite linkCompare serial line and satellite link
First, both paths idle; serial line have much less delayFirst, both paths idle; serial line have much less delay Then, traffic quickly overloads low capacity lineThen, traffic quickly overloads low capacity line Satellite delay will be less; change to itSatellite delay will be less; change to it
High capacity; load not significantly change delayHigh capacity; load not significantly change delay But, unloaded serial line will now become attractiveBut, unloaded serial line will now become attractive Routing will change again and the cycle will continueRouting will change again and the cycle will continue
Oscillations do occur in practiceOscillations do occur in practiceDifficult to manageDifficult to manage
Combining RIP, Hello, and BGPCombining RIP, Hello, and BGP
Single router may use multiple protocolsSingle router may use multiple protocols Interior Gateway ProtocolInterior Gateway Protocol
Gather routing information within ASGather routing information within AS Exterior Gateway ProtocolExterior Gateway Protocol
Advertise routes to other ASsAdvertise routes to other ASs Should be easy to combine the twoShould be easy to combine the two
Technical and political obstacles existTechnical and political obstacles exist
IGP protocols are routing protocolsIGP protocols are routing protocols RIP and HELLO used to update routing tablesRIP and HELLO used to update routing tables Get info from other routers inside ASGet info from other routers inside AS routedrouted implements RIP implements RIP
Advertises information from local routing tableAdvertises information from local routing table
Updates local table when it receives updatesUpdates local table when it receives updates
RIP RIP truststrusts routers within the AS to send correct data routers within the AS to send correct data
Exterior protocols (BGP) do not trust routersExterior protocols (BGP) do not trust routers Do not advertise all possible routes in local tableDo not advertise all possible routes in local table
Keep database of Keep database of reachability reachability
Apply policy constraints when sending/receiving infoApply policy constraints when sending/receiving info
Ignoring policy constraints can make some Ignoring policy constraints can make some parts of the internet unreachableparts of the internet unreachable
Example:Example:Suppose router running RIPSuppose router running RIP
Propagates route to Purdue; actually has no routePropagates route to Purdue; actually has no route
Other RIP routers will accept and updateOther RIP routers will accept and update
Will pass Purdue traffic to the erroneous routerWill pass Purdue traffic to the erroneous router
Problem if EGP protocol not have policy constraintsProblem if EGP protocol not have policy constraints Border router pass illegal route to other ASsBorder router pass illegal route to other ASs Purdue may become unreachable for parts of the internetPurdue may become unreachable for parts of the internet
Gated: Inter-AS CommunicationGated: Inter-AS Communicationgatedgated Interface between autonomous systemsInterface between autonomous systems Understands multiple protocolsUnderstands multiple protocols
Both IGP’s and BGPBoth IGP’s and BGP Ensures policy constraints are honoredEnsures policy constraints are honored
Can accept RIP msgs and modify local table (Can accept RIP msgs and modify local table (routedrouted))
Can advertise routes from within AS using BGPCan advertise routes from within AS using BGP Has rules on which networks it may & may not advertiseHas rules on which networks it may & may not advertise Also has rules on how to report distances to those networksAlso has rules on how to report distances to those networks
Links IGP with BGPLinks IGP with BGP
Open SPF Protocol (OSPF)Open SPF Protocol (OSPF)
Chapter 13: link state algorithmChapter 13: link state algorithm Uses SPF to compute shortest pathsUses SPF to compute shortest paths Scales better than distance-vector algorithmsScales better than distance-vector algorithms
OSPF is an IGP using link state algorithmOSPF is an IGP using link state algorithm Designed by Internet Engineering Task ForceDesigned by Internet Engineering Task Force To encourage adoption of link state technologyTo encourage adoption of link state technology Tackles several ambitious goalsTackles several ambitious goals
Open standardOpen standard Anyone can implement without license feesAnyone can implement without license fees
Includes Includes type of service routingtype of service routing Have multiple routes for a given destinationHave multiple routes for a given destination Choose by TOS field in IP headerChoose by TOS field in IP header OSPF first among TCP/IP protocols to have thisOSPF first among TCP/IP protocols to have this
Provides Provides load balancingload balancing Distributes traffic over multiple, same cost routesDistributes traffic over multiple, same cost routes
Can partition routers and networks into Can partition routers and networks into areasareas Permits growth; makes management easierPermits growth; makes management easier
Allows exchanges to be Allows exchanges to be authenticatedauthenticated Variety of authentication schemesVariety of authentication schemes
Supports host-specific, subnet-specific, and Supports host-specific, subnet-specific, and classless routesclassless routes
Accommodates multi-access nets (Ethernet)Accommodates multi-access nets (Ethernet)
Can describe network via virtual networkCan describe network via virtual network Abstracts away from details of physical Abstracts away from details of physical
connectionsconnections Provides flexibility for managersProvides flexibility for managers
Allows routers to exchange routing info Allows routers to exchange routing info learned from external siteslearned from external sites Distinguishes where information came fromDistinguishes where information came from
TypeType MeaningMeaning
11 Hello (used to test reachability)Hello (used to test reachability)
22 Database description (topology)Database description (topology)
33 Link status requestLink status request
44 Link status updateLink status update
55 Link status acknowledgementLink status acknowledgement
Figure 15.7Figure 15.7
Figure 15.8Figure 15.8
Routers exchange database description msgsRouters exchange database description msgs Used to initialize network topology databaseUsed to initialize network topology database During exchange:During exchange:
One router is master; other is slaveOne router is master; other is slave
Slave acknowledges each description with a responseSlave acknowledges each description with a response Topology database may be largeTopology database may be large
Can divide into several messages using Can divide into several messages using II and and MM bits bits
II = 1: is initial message = 1: is initial message
MM = 1: additional messages follow = 1: additional messages follow
BitBit S S indicates if sent by master (1) or slave (0) indicates if sent by master (1) or slave (0)
Sequence numbers used to make sure all receivedSequence numbers used to make sure all received
Link TypeLink Type MeaningMeaning
11 Router linkRouter link
22 Network linkNetwork link
33 Summary link (IP network)Summary link (IP network)
44 Summary link (link to border router)Summary link (link to border router)
55 External link (link to another site)External link (link to another site)
Figure 15.9Figure 15.9
Link status request messageLink status request message After exchanging DB descriptions, router may After exchanging DB descriptions, router may
discover parts of its DB are out of datediscover parts of its DB are out of dateRequests neighbor to send updateRequests neighbor to send update
Lists specific links it wants info aboutLists specific links it wants info about
Neighbor responds with most current information Neighbor responds with most current information about those linksabout those links
Figure 15.10Figure 15.10
Links from router to:Links from router to:
- given area- given area
- specific network- specific network
- single, subnetted IP network- single, subnetted IP network
- networks at other sites- networks at other sites
Figure 15.11Figure 15.11
Figure 15.12Figure 15.12
Routing with Partial InformationRouting with Partial Information
Hosts can have partial informationHosts can have partial information Rely on routersRely on routers
Not all routers have complete informationNot all routers have complete information Usually single router in AS connects to othersUsually single router in AS connects to others Suppose site connects to global InternetSuppose site connects to global Internet
At least one router must have connection to an ISPAt least one router must have connection to an ISP
Routers inside AS know all destinations withinRouters inside AS know all destinations within Have default route to send all traffic to the ISPHave default route to send all traffic to the ISP
Examine routing tablesExamine routing tables Routers at center of Internet know allRouters at center of Internet know all
Have complete set of routes to all destinationsHave complete set of routes to all destinations
Learn from the routing arbiter systemLearn from the routing arbiter system
They do They do notnot use default routing use default routing
If destination address not in their table:If destination address not in their table: Either address is not valid, orEither address is not valid, or Address is valid but currently unreachableAddress is valid but currently unreachable
Routers beyond those at center do not know allRouters beyond those at center do not know allHave an incomplete set of routesHave an incomplete set of routes
Rely on default routes for addresses they do not haveRely on default routes for addresses they do not have
Using default routes for most routers has Using default routes for most routers has consequencesconsequences Local routing errors can go undetectedLocal routing errors can go undetected
Machine in AS routes packet to external ASMachine in AS routes packet to external AS
Suppose should have gone to local routerSuppose should have gone to local router
External system will route it backExternal system will route it back
Preserves connectivity even if routing incorrectPreserves connectivity even if routing incorrect
Very bad for a WANVery bad for a WAN Routing update messages will be smallerRouting update messages will be smaller
A good thing!A good thing!
SummarySummaryMust have automated routing proceduresMust have automated routing proceduresIGPsIGPs Used by 2 routers under control of single mgrUsed by 2 routers under control of single mgr Uses either DV or link state algorithm (SPF)Uses either DV or link state algorithm (SPF)
RIP: DV implemented by RIP: DV implemented by routedrouted Uses split horizon, hold-down and poison reverseUses split horizon, hold-down and poison reverse
Hello: obsolete; uses delay versus hop countHello: obsolete; uses delay versus hop countOSPF: uses link status algorithmOSPF: uses link status algorithm
gatedgated Interface between IGP (RIP) and EGP (BGP)Interface between IGP (RIP) and EGP (BGP)