nennatwarkinu

IMO) anmd P'IM SANM in a t | Wito| and | 9 ) | are botli very readalle ino
d tim tn muvant wih intewg litoal deail
e han een a o of wok developig aloniolmn thar n e ed by outers
o fant lvokup v ' auldovNe, (Recall thar the llemn in that lie outer needs to
math he longest vetis in the lorwanding talble.) TATRIC IA tees are one of the ine
alguiuhm applied tv this puoblem |Muro|. Mone reen work iepoted in |DBS"97|.
WTT)1. |LS9N|, and |8VSW"N|. lor an vervicw l huow hese algorithms can be
ued io buikl hugh spel
a nten,
|'a N| se l'atnlge et al.
Muliproool label
switching und the related protocols that led developmen is

ae deilhed in Chandranmenom and Varghese |V951, Reklhter al. |RDR'97|. .and et

Davic anl Rekhuer
|DRoOj. T'he later velerene deseuibes many applhcaon of MILS
sIh as rathe engineering. tast recovery lrom network lailures, and viruual
wwrks. RROG) private inet
provides the specilication of MPLS/BGP VI'Ns, a lormof layer VPN
that can be
provided over MPIS networlks.
Finally, recommend he lollowing live velerences:
we

http://www.ietf.org: The
ETE home page, from wlhih you can get REC,
intcnet dralis, and working group charteIs.
http://www.isoc.org/internet/history/:
history, including
Icnet Acollection of links related to In-
articles written mome
builh the lnienet. by tlhe original researchcrs who

E XERCI S E S
1 What aspect of 1 addresses
makes it necessury to have
work interface, rather one address
than just one host? per net-
does IP tolerate per In light of your answer, why
addresses? point-to-point interfaces that have
nonunique addresses or no
2
Why does the Offset field in the
(Hin: Recall that 1P header measure the oftset
the Offset field is 13 bits
in 8-byte units?
3 Some
long.)
sgnalling cirors can cause entire
en by all 0s ranges ot bius in
all Is.
or
Suppose all the bits in the packet a
to be overwrit
checksun are packet
overwritten. Could packet with all Os orincluding the lutenet
acke' Will he a
Internet checksum catch all Is be
that crror Why or why legal IP'v4
a

not?
upposeA TCP
messpge that contains
ieader
passed to IP lor delivery 2,048 bytes of data
is and 20
bytes of T'CP
Irom the source across two networks ot the
host to a lnternet (i.e.,
ouer the
to
destination hosSt). T he
iust network
 Exerolea 301

uwY 4 byte headers and has an MTU t 1,024 bytes: tlhe seconmd usew lbyte
hcaders wih an AUrU' ot S12 bytes. Fach networks MTgves the ue
the langest l 'datag1ann that can be canied in a ink layer Ihame, Give the sises
and ottsets ot the soquewe of tragments delivered to the network layer at the
destinatiou host. Assume all 1P headers are 20 bytes.

5Yhth MTU is the smallest MTU of any link on the cuwrent path (route) be
c n w hosts. Assume we could discover the Patlh MTU of the path used
in the prevwns exereisc. and that we use this valuc as the MTU lor all thhe path
segments. Give the sires and oftsets of the sequenee of fragments delivered to
the network layer at the destination host.

6 Suppose an ll' packet is tragmented into 10 fragments, cach with a I% (inde-

pendent) probability of loss. lo a reasonable approximation, this means there
is a 10% chance of losing the whole packet duc to loss of a fragment. What
is the probability of net loss of the whole packet if the packet is transmitted
tWIce.

(a) Assuming all fragments received must have been part of the same trans-
mission?

b) Assuming any given fragment may have been part of either transmission?

(c)Explain how use ofthe ldent field might be applicable here

7 Suppose the fragments of Figure 4.5(b) all pass through another router onto
a link with an MTU of 380 bytes, not counting the link header. Show the
agments produced. If the packet were originally tragmented for this MTU,
how many fragments would be produced?

8 What is the maximum bandwidth at which an IP host

can send
576-byte pack-
ets without having the ldent field wrap around within 60 seconds? Suppose
IP's maximum segment litetime (MSL) 1s 60 scconds, that is, delayed pack
ets can arrive up to 60 seconds late but no later. What might happen if this
bandwidth were exceeded?

9 ATM AAL3/4 uses fields Btag/Etag, BASize/Len, Type, SEQ, MID,

Length, and CRC-10 to implement fragmentation into cells. IPvá uses
Ident, Offset, and the M bit in Flags, among others. What
is the IP analog.
if any, for each AAL3/4 held: Does each IP
field listed here have an AAL3/4
analog? How well do these helds correspond?
62 4ermetworking

done at the endpoint. rather

10 h do ou think IP4 has tragment reassembly
IPv6 abandoned fragmentation
han ar the nexr rourter Whv do vou think
encirev Hinc Think ibout the ditferences between IP-layer tragmentation
and ink-Ter tramencaion.)

11 Hzumg ARP ble entries time out after 10 to 15 minutes is an attempt at a

sozbie compromse. Describe the problems that can occur if the timeout
aue s coo Sml oc too karg

12 Pueiv uses 1-bir addresses. Ifwe could redesign IP to use the 6-byte
SLAC dress instezd of the 32-bt address. would we be able to eliminate the
e e oe RP: Expiin whr or why noc.

13 Supse hosts A nd B here been assigned the same IP address on the same
Eemer cr wich ARP s used. B stars up after A. What will happen to
s isting coeneions Ezplain how "self-ARP (querving the network on
- r odei P ddress) might help with this
own
problem.
14 oese in
Pimpiemencation adheres literally to the following algorithm on
or atz acker P desined tor iP address D

ree acarees fer Dis ir

ARP cacre
serc

erc or r RP aLery for D

comes cack)
e P
aver receives a burst of packers destined for D, how
igehnm wzste tesources unnecessar1ly: might this
enc zn
moroved versuon
ppes e
sumpir drop P
zls
ier
sending our query, when cache lookup
a
t nus behave Some early ARP implementations
allegedly
15 For er
en in
Figure +50. give global distance-v
ector tables like

he distarces to its immediate neigh bors.

meiare 1eigr
he
intorrmation it nad in the
preceding step to its
 Exercises 363

D 2

Figure 4.50 Network for Exercises 15, 17, and 20.

2 1
A B E

D C 3
F

Figure 4.51 Network for Exercise 16.

second time.
(c) Step (6) happens a
distance-vector tables like
16 For the network given in Figure 4.51, give global
those of Tables 4.5 and 4.8 when

(a) Each node knows only the distances to its immediate neighbors.
it had in the preceding step to its
(b) Each node has reported the information
immediate neighbors.

(c) Step (b) happens a second úme.

17 For the nerwork given in Figure 4.50, show how the link-state algorithm builds
the routing table for node D.

18 Suppose we have the forwarding tables shown in Table 4. 12 for nodes A and
E in a network where all links have cost 1. Give a diagram of the smallest
network consistent with these tables.
 364 4 Internetworking

F
A
Node Cost NextHop
Node Cost NextHop
A 3 E
B B

B B

B D E

F 3 D E E

Table 4.12 Forwarding tables for Exercise 18.

Node Cost NextHop Node Cost NextHop

B B

C B

D C C
E D C
F C E E
Table 4.13 Forwarding tables for Exercise 19.

19 Suppose we have the forwarding tables shown in Table 4.13 for nodes A and
F in
network where all links have
a
cost 1. Give a
diagram of the smallest
nerwork consistent with these tables.

20 For the network in

Figure 4.50, suppose the forwarding tables
lished are all estab-
as in Exercise 15 and then
the C-E link fails. Give
(a) The tables of A, B, D, andF after C and E
have reported the news.
(b) The tables of A and D
after their next mutual
(c) The table
exchange.
of
C

afier A
exchanges with it
21 Suppose a router has built up the
router can deliver routing table shown in Table 4.14. The
packets direcly over interfaces 0 and 1, or it can
packets to routers R2, R3, R4. torward
addressed each of the
or
Describe what the router does
to
with a packet
following destinations:
 Exercises 365

SubnetNumber SubnetMask NextHop

128.96.39.0 255.255.255.128 Interface 0

128.96.39.128 255.255.255.128 Interface 1

128.96.40.0 255.255.255.128 R2

192.4.153.0 255.255.255.192 R3

(default) R4

Table 4.14 Routing table for Exercise 21.

SubnetNumber SubnetMask NextHop

255.255.254.0 Interface 0
128.96.170.0

128.96.168.0
255.255.254.0 Interface 1

128.96.166.0 255.255.254.0 R2
128.96.1644.0 255.255.252.0 R3
R4
default)
Table 4.15 Routing table for Exercise 22.

(a) 128.96.39.10.

(b) 128.96.40.12.
y
(c) 128.96.40.151. ra

(d) 192.4.153.17.

(e) 192.4.153.90.

routing table shown in Table 4.15. The

router has built up the
22 Suppose a
intertaces 0 and 1, or it can forward
router can deliver packets directly over

R2, R3, or R4. Assume

the router does the longest prefix
to routers
packets each of the
router does with a packet addressed
to
match. Describe what the
following destinations:

(a) 128.96.171.92.

b) 128.96.167.151.
(c) 128.96.163.151.
366 4 Internetworking

Figure 4.52 Simple network for Exercise 23.

(d) 128.96.169.192.
(e)128.96.165.121.

23 Consider the
simple network in Figure 4.52, in which A and B exchange
distance-vector routing information. All links have cost 1.
link fails. Suppose the A--E

(a) Give a sequence of routing table

tween A and B.
updates that leads to a
routing loop be

(b) Estimate the probability of the scenario in (a), assuming A and B send out
routing updates at random times, each at the same
average rate.
(c)Estimate the probability of loop
report within 1 second of
forming if A broadcasts an updated
a

discovering
every 60 seconds uniformly.
the A-E failure, and B
broadcasts
24 Consider the situation
of involving the creation of a routing loop in the
Figure 4.14 when the A-E link nerwork
dates among A, B, and goes down. List all of
C, pertaining to sequences table
Assume that table destination E, that lead to the up-
updates are done one at a time, that the loop.
nique is observed by all
E's participants, and that A sends itssplit-horizon tech
unreachability to B before C. You may initial report of
changes. ignore updates that don't result in
25
Suppose a set of routers all use the
under what
circumstances it makes split-horizon technique;
difference if they use
we consider here
addition. a
poison reverse in

(a) Show that

poison reverse
makes no difference in
ing loop in the two the evolution of the
hosts involved use examples described in Section 4.2.2, rout-
split horizon. given that the
(b) Suppose split-horizon routers A
they forward tratfic for a
and B somehow
reach a state in
how this situation will given destination X toward which
evolve with and each other. Describe
without the use of
poison reverse.
 Exercises 367

and E B
B
E

Networks for Exercise 26.

Figure 4.53

Network for Exercise 27.

Figure 4.54
state as in (b),
even

that leads A and B to a looped

(c) Give a sequence of events and A c o n n e c t through very
a

r e v e r s e is used.
(Hint: Suppose B
if poison
X through a third node,
C, and simultaneously
each reach
slow link. They
advertise their r o u t e s to each other.)
hosts
distance-vector loop-avoidance technique, whereby
another a chance to
26 Hold down is n e w s has had
of time until link failure
for period 1
where all links have cost
a
ignore updates nerworks in Figure 4.53,
Consider the
breaks and B reports
propagate.
10. Suppose that the E-A link
E-D with cost
afterwards (this is the false
route,
except
E route to A immediately
its loop-forming hold-down interpretation,
and use this to de-
details of a
via A). Specify the in both networks. To
what extent can
of the routing loop
scribe the evolution without delaying the discov-
in the EAB nerwork
hold down prevent
the loop
alternative route in
the EABD network?
ery of
the
the
4.54, using link-state routing. Suppose
the network in Figure
2 7Consider in sequence:
the following then
occur

B-F link fails, and

side withaconnection to G.
added to the right
(a) Node H is
connection to C.
added to the lett side with a

(b) Node D is
added.
link D-A is
(c) A new

is restored. Describe what link-state packets will flood

The failed B-F link
now

number at all nodes is 1, and

Assume that the initial sequence
back and forth.
 368 4 Internetworking

5D
2 2
5
A B

Figure 4.55 Network for Exercise 28.

6
3
3 E
Figure 4.56 Network for Exercise 29.

Figure 4.57 Network for Exercise 30.

that no
packets time out, and that
both ends of a link
number in their LSP for the same
use
that link, greater than sequence
used before. any sequence number
either
28 Give the
steps as in Table 4.9 in
routing database for node A in thethe forward search algorithm as it builds the
network shown in
29 Give the
steps in Table as
Figure 4.55.
4.9 in
routing database for node A in thethe forward search algorithm as it builds the
network shown in
30
Suppose that nodes in the network Figure 4.56.
State
routing, shown in
and C receives Figure 4.57 participate in
the A-B link is contradictory
down, but one from B LSPs: One from A arrives link
arives claiming the A-B claiming
(a) How link is up.
could this
happen?
(b) What should
C do? What
can C
Do not
expec?
assume that LSPs contain
any
synchronized timestamp.
282 4 Internetworking

5 B
10
A
11

Figure 4.18 Link-state routing: an example network.

Step Confimed Tentative Comments

(D.0-) Since D is the only new member of the confirmed list,
look at its LSP.
2 (D.0-) (B,11,B) D's LSP says we reach B
(C.2,C) is better than
can
through B at cost 11, which
anything else on either list, so
put it on Ten-
tative list; same for C.
3 (D,0,-) (B,11,B) Put lowest-cost member of
Tentative (C) onto
(C,2.C) firmed list. Next, examine LSP of Con
member (C). newly conhrmed
4 D,0,-) (B,5,C) Cost to reach B
(C,2,C) (A,12,C) through C is 5,
LSP tells us that we can reach A
so
replace (B,11,B). Cs
at cost 12.
(D,0,-) (A, 12,C) Move lowest-cost member of Tentative
(C,2.C) firmed, then (B) to Con-
(B,5,C) look at its LSP:

6 (D,0,-) (A,10,C) Since we can reach A at

(C.2.C) cost 5 through B, replace the Ten-
tative entry.
(B,5,C)
(D,0-) Move
(C2,C) lowest-cost member of
(B,5.C) firmed, and we are all done. Tentative (A) to Con
(A,10,C)
Table 4.9 Steps for
building routing table for node D
(Figure 4.18).
The link-state routing
algorithm has many nice
stabilize quickly, it does not generate n properties: It has been
proven to