Chapter 4 LinkState Routing

1
Chapter 4Link-State Routing

• Professor Rick Han
• University of Colorado at Boulder
• rhan_at_cs.colorado.edu

2
Announcements

• Should use sendto_(), not sendto()
• This weeks lectures online by Saturday
• HW4 online by Saturday
• Updated PA 2 (small mods)
• Server should exit cleanly
• New library void()
• recvfrom() at server instead of select()
• Next, link-state routing,

3
Recap of Previous Lecture

• Distance vector problems
• Bouncing effect
• Bad news travels slowly
• Counting to infinity
• Solutions
• Holddown
• Split horizon with poisoned reverse
• Bellman-Ford Equation
• D(i,j) min d(i,k) D(k,j) for all iltgtj
• k neighbors

4
Recap of Previous Lecture (2)

• Bellman-Ford Algorithm Ford Fulkerson
• D(i,j)h1 min d(i,k) D(k,j)h for all
  iltgtj, h0,1,
• k neighbors
• Distributed Bellman-Ford Algorithm Distance
  Vector
• Physically distribute the calculation of
  D(i,j)h to router i only, and
• Exchange the results of your D(i,j)h with
  neighboring routers at each iteration h

5
Routing Information Protocol (RIP)

• RIP is a specific realization of the distance
  vector or distributed Bellman-Ford routing
  algorithm
• Distance vectors are carried over UDP over IP
• RIP uses hop count as its shortest path metric,
  so d(i,j)1
• Distance vectors are sent every 30 seconds
• When a routing table changes, a router can send
  triggered updates to neighbors before 30 sec
• Can lead to network storms, so limit rate wait 5
  seconds between sending new routing update and
  the update that caused routing table to change

6
RIP is simple

• At each step, exchange distance vectors with each
  neighbor
• Update distance table with new distance vector,
  adding one (all link costs are one)
• Calculate minimum hop path to each destination by
  looking at minimum in the row

Distance table at B
A
B
Via Port/Link
C
7
RIP problems

• N-hop loops
• Metric of hop count doesnt capture Internet
  loading
• Doesnt scale well
• The neighbor information carried by distance
  vectors takes awhile to propagate to across of
  network
• If your network diameter is 20 hops, then it can
  take 20(30 seconds) 10 minutes to propagate
  all neighbor info and converge to stable
  routing tables
• How could we improve speed of convergence?
• Flood distance vectors, except theyre large!

8
Alternative Routing Strategy

• Instead, flood just your neighbor link costs and
  compute a shortest path tree
• First flood your neighbor link costs to entire
  network
• These are small packets
• Each node receives entire topology
• Second, each node computes the shortest path
  tree with itself at as the root

• This is called Link-State Routing

9
Alternative Routing Strategy (2)

• How is this shortest path tree calculated?
• Dijkstras Observation shortest path to nodes
  further from the root must go through a branch of
  the shortest path tree closer to the root

• Strategy calculating the shortest path tree from
  the root (B), add nodes to the tree
• Add node closest to root
• at each iteration

10
Dijkstras Shortest Path Algorithm

• Let N set of nodes in graph
• l(i,j) link cost between i,j ( infinity if not
  neighbors)
• SPT set of nodes in shortest path tree thus far
• S source node
• C(n) cost of path from
• S to node n

11
Dijkstras Shortest Path Algorithm (2)

• Initialize shortest path tree SPT S
• For each n not in SPT, C(n) l(s,n)
• While (SPTltgtN)
• SPT SPT U w such that C(w) is minimum for all
  w in (N-SPT)
• For each n in (N-SPT)
• C(n) MIN (C(n), C(w) l(w,n))

12
Dijkstras Shortest Path Algorithm (3)

• Initialize shortest path tree SPT B
• For each n not in SPT, C(n) l(s,n)
• C(E) 1, C(A) 3, C(C) 4, C(others)
  infinity
• Add closest node to the root SPT SPT U E
  since C(E) is minimum for all w not in SPT.
• No shorter path to E can ever be found via some
  other roundabout path.

• Shortest path tree SPT B, E

13
Dijkstras Shortest Path Algorithm (4)

• Recalculate C(n) MIN (C(n), C(E) l(E,n)) for
  all nodes n not yet in SPT
• C(A) MIN( C(A)3, 1 1) MIN(3,2) 2
• C(D) MIN( infinity, 1 1) 2
• C(F) MIN( infinity, 1 2) 3
• C(C ) MIN( 4, 1 infinity) 4

• Each new node in tree, could create a lower cost
  path, so redo costs

14
Dijkstras Shortest Path Algorithm (5)

• Loop again, select node with the lowest cost
  path
• C(A) 2, C(D) 2, C(F) 3, C(C ) 4
• SPT SPT U A B, E, A
• No shorter path can be found from B to A,
  regardless of any new nodes added to tree

• Recalc C(n) MIN (C(n), C(A) l(A,n)) for all
  n not yet in SPT
• C(D) MIN(2, 2inf) 2
• C(F) 3, C(C) 4

15
Dijkstras Shortest Path Algorithm (6)

• Continue to loop, adding lowest cost node at each
  step and updating costs
• SPT crawls outward
• Remember to store the links in SPT as they are
  added (each nodes predecessor is stored)
• Each node has to store the entire topology, or
  database of link costs

16
Link-State Routing Reliable Flooding Dijkstra
SPT

• Start condition
• Each node assumed to know state of links to its
  neighbors
• Step 1
• Each node broadcasts its state to all other nodes
• Reliable flooding mechanism
• Step 2
• Each node locally computes shortest paths to all
  other nodes from global state
• Dijkstras shortest path tree (SPT) algorithm

17
Link State Packets (LSPs)

• Periodically, each node creates a link state
  packet containing
• Node ID
• List of neighbors and link cost
• Sequence number
• Needed to avoid stale information from flood
• Time to live (TTL)
• Node outputs LSP on all its links

18
Reliable Flooding of LSPs

• When node J receives LSP from node K
• If LSP is the most recent LSP from K that J has
  seen so far, J saves it in database and forwards
  a copy on all links except link LSP was received
  on
• Otherwise, discard LSP
• How to tell more recent
• Use sequence numbers
• Same method as sliding window protocols
• Propagates most recent info avoids loops
• Not truly reliable in that LSPs can be lost
• Reliable due to periodic flooding
  retransmission
• Transient loops can still occur

19
OSPF Open Shortest Path First

• A particular realization of link-state routing,
  used for intra-domain routing in the Internet
• Additional support for
• Authentication of routing updates
• Support for broadcast networks
• Different cost metrics
• Periodic and event-triggered flooding of LSP
  routing updates, like RIP

20
Link State vs. Distance Vector

• Routing update size
• LS small, contain only neighbors link costs
• DV potentially long distance vectors (length N
  for N nodes in network)
• Routing update communication overhead
• LS flood to all nodes, overhead is O(NE), where
  N is nodes, and E is edges or links
• In DV, send distance vector only to neighbors

21
Link State vs. Distance Vector(2)

• Convergence speed
• DV at each iteration, send to neighbor and
  recalculate
• takes awhile to propagate changes to rest of
  network
• Iterations are periodic, hence slow faster when
  triggered
• LS faster dont need to recalculate LSPs
  before forwarding
• may be key reason that LS beat out DV in
  intra-domain routing

22
Link State vs. Distance Vector (3)

• Space requirements
• LS maintains entire topology in a link database
• DV maintains only neighbor state
• If each of N routers has K neighbors,
• LS O(NK) memory requirement
• DV O(NK) also

23
Link State vs. Distance Vector (4)

• Complexity (initializing from scratch)
• DV O(NKDiameter)
• for each of N rows in distance table, find min of
  (dikDkj) over K neighbors, iterate until get
  info via DVs of furthest nodes (a diameter away)
• If sorted list kept, reduce complexity to
  O(Nlog(K)Diam.)
• LS O(N(N-1)/2)) O(N2)
• First iteration, find min cost node from N-1
  nodes not in SPT for 2nd iteration, find min
  cost node from N-2 nodes not in SPT,
• If a sorted list kept, reduce complexity to
  O(Nlog(N))
• After convergence, new routing updates may only
  spur partial recalculations for both DV LS

24
Link State vs. Distance Vector (5)

• Robustness
• Both LS and DV can be completely disabled by a
  single router advertising false/corrupt LSP or DV
• LS can flood false/corrupt LSPs to all routers
• DV can advertise false paths/costs to all
  neighbors
• In ARPANET, malfunctioning routers have
  advertised zero cost, creating a black hole
• In 1997, a bad router in a small ISP advertised a
  false cost, became flooded with traffic,
  disconnecting ISPs from most U.S. backbone
  providers for 3 hours

25
Link State vs. Distance Vector (5)

• Bottom line
• no clear winner in terms of complexity, space,
  robustness,
• but LS is favored in the intra-domain Internet
  due to faster convergence

26
Link-State Cost Metric

• Choice of link cost defines traffic load
• Low cost high probability link belongs to SPT
  and will attract traffic, which increases cost
• Choices for metric
• hop count
• queueing delay
• Transmission delay
• Propagation delay

27
Link-State Cost Metric (2)

• Static metrics (e.g. hop count or another fixed
  cost)
• Less overhead than dynamic metrics
• flood LSPs once initially,
• Thereafter, flood LSPs only if link fails
• Dont have to deal with staleness
• Dynamic metrics can be out of date by the time
  they arrive at a distant router

28
Link-State Cost Metric (3)

• Dynamic metrics take into account
• Changes in link delay (due to congestion)
• Variations in link capacity (wireless BW)
• Dynamic metrics should
• Avoid oscillations
• low cost attracts traffic gt increase cost gt
  less traffic gt low cost gt increase traffic gt
  increase cost
• Achieve good network utilization
• Use link BW efficiently
• Limit overhead from flooding LSPs
• Respond quickly, to avoid performing Dijkstra on
  stale info

29
Original LS ARPANET Metric

• Cost proportional to queue size
• Instantaneous queue length as delay estimator
• Problems
• Did not take into account link speed
• Did not take into account propagation delay of a
  link
• Poor indicator of expected delay due to rapid
  fluctuations
• Moves packets toward smallest queue, rather than
  to destination
• Does not achieve good network utilization

30
New LS ARPANET Metric

• Delay (depart time - arrival time)
  transmission time link propagation delay
• (Depart time - arrival time) captures queuing
• Transmission time captures link capacity
• Link propagation delay captures the physical
  length of the link
• Measurements averaged over 10 seconds
• Update sent if difference gt threshold, or every
  50 seconds
• Achieves better network utilization

31
Problems With New Metric

• Works well for light to moderate load
• Static values dominate
• Oscillates under heavy load
• Queuing dominates
• Congested link advertising high cost pushes
  traffic away gt some links temporarily
  underutilized during heavy load 50 given 2
  links between 2 nodes
• Range is too wide
• 9.6 Kbps highly loaded link can appear 127 times
  costlier than 56 Kbps lightly loaded link
• Can make a 127-hop path look better than 1-hop
• Satellite links penalized, though theyd better
  suit playback video (high BW, non-delay
  sensitive)