Hierarchical Routing

woonsocketpoliticalΔίκτυα και Επικοινωνίες

28 Οκτ 2013 (πριν από 3 χρόνια και 10 μήνες)

111 εμφανίσεις

Hierarchical Routing
EECS 228
Abhay Parekh
parekh@eecs.berkeley.edu

Hierarchical Routing
Is a natural way for routing
5 7
to scale
4
4
RIP
8
Size
6
6
Network Administration
Governance
11
2
10
2
Inter Domain
Exploits address
OSPF
Routing
aggregation and allocation
3
1 3 13 13
12
Allows multiple metrics at
IGRP
different levels of the
hierarchy
October 16, 2002 Abhay K. Parekh: Topics in Routing 2

Why is hierarchical routing important?
The internet is an interconnection of unequal
networks
Interconnection arrangements drive
the competitive landscape
the robustness of the network
end-to-end performance
Interconnection is central to all large networks
Voice
Data
Wireless
Cable
October 16, 2002 Abhay K. Parekh: Topics in Routing 3


Why are there so many players?
www.thelist.com
How many ISP s in the 415 area code?
That start with A-C: about 200
Just DSL that start with A-C: about 80
In the telephone network
How many independent telephone companies in
1894-1902 in the US?
3039 commercial companies, 979 co-operatives
By controlling interconnection Bell got rid of them
Interconnection is now regulated (CLECs)
October 16, 2002 Abhay K. Parekh: Topics in Routing 4


What is an Interconnection
Access to sites reachable via routing and transport facilities
But could also include:
Wire
SLA + Lease
Space
Size
Space
Access to OSS
Dispute Resolution Process
Term
October 16, 2002 Abhay K. Parekh: Topics in Routing 5Interconnections occur at many levels
B
5 7
4
RIP
4
8
6 6
11
2
2 10
Inter Domain
OSPF
Routing
3
1 3 13 13
12
IGRP
C
A
October 16, 2002 Abhay K. Parekh: Topics in Routing 6Interconnections occur at many levels
B
5 7
A.1 D.2
4
RIP
4
8
6 6
11
2
2 10
OSPF
3
D.1 D.3
1 3 13 13
12
IGRP
C
A
October 16, 2002 Abhay K. Parekh: Topics in Routing 7Interconnections occur at many levels
B
5 7
A.1 D.2 E.1
4
RIP
4
8
6 6
11
E.2
2
2 10
OSPF
3
D.1 D.3
1 3 13 13
12
E.3
IGRP
C
A
October 16, 2002 Abhay K. Parekh: Topics in Routing 8Interconnections occur at many levels
B
5 7
A.1 D.2 E.1
4
RIP
4
8
6 6
11
E.2
2
2 10
OSPF
3
D.1 D.3
1 3 13 13
12
E.3
IGRP
C
A
October 16, 2002 Abhay K. Parekh: Topics in Routing 9Interconnections occur at many levels
E
5 7
A.1 D.2 E.1
11
E.2
D
D.1 D.3
1 13 13
12
E.3
October 16, 2002 Abhay K. Parekh: Topics in Routing 10
Examples of overlaid interconnecting
networks
IP over ATM
Multicast over IP
DSL over POTS
IP over CATV
Etc.,
Each involves routing
October 16, 2002 Abhay K. Parekh: Topics in Routing 11

Two ways to interconnect IP Networks…
Peering
The business relationship whereby ISPs
reciprocally provide to each other connectivity to
each others transit customers
Transit
The business relationship whereby one ISP
provides (usually sells) access to all destinations
in it s routing table
William B. Norton, “Internet Service Providers and Peering”
October 16, 2002 Abhay K. Parekh: Topics in Routing 12Figures from
Peering and Transit
William B. Norton, “Internet Service Providers and
Peering”
October 16, 2002 Abhay K. Parekh: Topics in Routing 13Benefits of Transit v/s Peering
William B. Norton, “Internet Service Providers and Peering”
October 16, 2002 Abhay K. Parekh: Topics in Routing 14Moving from Transit to Peering
William B. Norton, “Internet Service Providers and Peering”
October 16, 2002 Abhay K. Parekh: Topics in Routing 15


Peering Attributes
Bandwidth Pricing: Everything you can think of
Traffic may be asymmetric (web servers)
Clout may vary
Some existing and suggested methods
Zero Charge (Bill and Keep)
Average Cost
Fully distributed cost pricing
Ramsey Pricing
Wholesale Pricing
Marginal Cost Pricing
Bandwidth is undifferentiated (can t peer for video quality bw)
Connection Method
Direct Connection
Internet Exchange
October 16, 2002 Abhay K. Parekh: Topics in Routing 16
Undifferentiated Network here to stay?
Undifferentiated Network here to stay?
Internet Infrastructure provides
undifferentiated service
No Business
Model Cop-out
No way to charge, peer or deliver
high speed/quality sensitive
applications
More capacity is thrown at the undifferentiated network, and emphasis
continues on speeding up the internet , but this just speeds up existing
applications
No future for internet media or other bandwidth intensive applications
No future for significant high speed access penetration
These are huge lost opportunities!!
October 16, 2002 Abhay K. Parekh: Topics in Routing 17

Name of the Game: Reachabilty
BGP is the way by which ISPs co-operate on
reachability
Routing efficiency and performance is important, but
not essential
E.g. Path Vector uses many messages
Power of BGP is that it can express many different
ISP routing policies without exposing internal
network statistics such as load and topology
Tremendous growth in the last 10 years
October 16, 2002 Abhay K. Parekh: Topics in Routing 18October 16, 2002 Abhay K. Parekh: Topics in Routing 19


Skitter Legend
Plot the AS based on polar co-ordinates (r,θ):
r = 1- log (As degree +1 / Max Degree+1)
Higher the degree lower the radius
Θ = longitude of AS headquarters
October 16, 2002 Abhay K. Parekh: Topics in Routing 20
4/1-4/16 2002
1,224,733 IP addresses,
2,093,194 IP links,
932,000 destinations,
70% of globally routable network prefixes;
10,999 ASes (84% of ASes),
34,209 peering sessions
October 16, 2002 Abhay K. Parekh: Topics in Routing 21October 16, 2002 Abhay K. Parekh: Topics in Routing 22

BGP
Runs over TCP port 179
One Border Routers can be involved in multiple
sessions
Border Routers
from the same AS speak IBGP
from different AS s speak EBGP
EBGP and IBGP are essentially the same protocol
IBGP can only propagate routes it has learned directly from
its EBGP neighbors
All routers in the same AS form an IBGP mesh
Important to keep IBGP and EBGP in sync
October 16, 2002 Abhay K. Parekh: Topics in Routing 23
Four message types
Open: Session establishment id exchange
Notification: exception driven information
Keep Alive: soft state
Update: path vector information
October 16, 2002 Abhay K. Parekh: Topics in Routing 24
Update Message
Withdrawn Routes: No
Infeasible Route Length
longer valid
Withdrawn Routes
Attributes: Path Vector,
(variable)
weights and other
Total Path Attribute Len
information about each of
Path Attributes
the destinations
(variable)
Length|Prefix
<length,prefix>: CIDR
notation for the destination
<length,prefix>
.
.
October 16, 2002 Abhay K. Parekh: Topics in Routing 25


Classless Inter-domain Routing Addresses
32 bits in the address divided into 4 8-bit parts, A.B.C.D
Each part takes value 0,1,2, ,255
E.g. 128.23.9.0
Specify a range of addresses by a prefix: X/Y
The prefix common to the entire range is the first Y bits of X.
X: The first address in the range has prefix X
32-Y
Y: 2 addresses in the range
Example 128.5.10/23
Common prefix is 23 bits: 01000000 00000101 0000101
9
Number of addresses: 2 = 512
Prefix aggregation
128.5.10/24 and 128.5.11/24 gives 128.5.10/23
Addresses allocated by central authority: IANA
Routers match to longest prefix
October 16, 2002 Abhay K. Parekh: Topics in Routing 26


Advertising a prefix
One router telling another one
The prefix
IP address of the next hop
Path list of AS s that the announcement has
passed through
Since announcement propagates from destination, this
yields the path
No refresh messages required
The announcing router will follow the path
itself
October 16, 2002 Abhay K. Parekh: Topics in Routing 27Example
b
B
5 7
4
RIP
4
8
IBGP
6 6
11
2
2 10
a
IBGP
OSPF
BGP Session
3
c
1 3 13 13
12
b via A
IGRP
C
A
Announce a
October 16, 2002 Abhay K. Parekh: Topics in Routing 28
Multihoming
Two or more
B
5 7
interdomain
4
RIP connections between
4
8
IBGP
the same AS s
6 6
Two or more
interdomain
2
2
connections between a
a
IBGP
customer and ISPs
3
1 3
IGRP
A
October 16, 2002 Abhay K. Parekh: Topics in Routing 29
Multiexit Discriminators (MEDs)
Way for one AS to influence
B
routing decisions of another
5 7
AS
4
RIP
4
AS_A wants to tell AS_B that
8
network a is closer to router 2
IBGP
than to router 3
6 6
Router 2 advertises a smaller
MED value for a than Router 3
2
AS_B prefers the path to a that
2
does not go through 6 and 3
a
IBGP
AS_B does not propagate
3
1 3
MEDS from AS_A any further
IGRP
A
October 16, 2002 Abhay K. Parekh: Topics in Routing 30

Local Preference (for IBGP)
Similar to MEDs but rather than being part of
the EBGP announcement, is a way for IBGP
within an AS to prefer one path over another
for the same prefix
Example
Choose the slower path when the prefix is to a
competitor s network
October 16, 2002 Abhay K. Parekh: Topics in Routing 31

Multihoming
Advantages
Redundancy
Load Sharing
Lowest Cost Routing
Lowest Latency Routing
Disadvantages
Aggregation (if one customer has two ISPs)
October 16, 2002 Abhay K. Parekh: Topics in Routing 32


BGP Policies
Multiple ways to implement a policy
Decide not propagate advertisements
I m not carrying your traffic
Use MEDs to improve performance
Decide not to consider MEDs but use shortest hop
Hot potato routing
Prepend own AS# multiple times to fool BGP into
not thinking AS further away
Many others
October 16, 2002 Abhay K. Parekh: Topics in Routing 33



BGP and Performance
BGP isnt designed for policy routing not performance
Hot Potato routing is most common but suboptimal
Performance isn t the greatest
20% of internet paths inflated by at least 5 router hops
Very susceptible to router misconfiguration
Blackholes: announce a route you cannot reach
October 1997 one router brought down the internet for 2 hours
Flood update messages (don t store routes, but keep asking your
neighbors to clue you in). 3-5 million useless withdrawals!
In principle, BGP could diverge
Various solutions proposed to limit the set of allowable policies
Focuses on avoiding policy cycles
October 16, 2002 Abhay K. Parekh: Topics in Routing 34
BGP Updates (Labovitz)
Most updates were bogus withdrawals
This was to a large extent due to bad
implementations
October 16, 2002 Abhay K. Parekh: Topics in Routing 35
Non-Withdrawal Updates (Labovitz)
AADup is the advertising of a route that was
just withdrawn
October 16, 2002 Abhay K. Parekh: Topics in Routing 36
BGP Bouncing Problem (Labovitz et al)
Initially direct route
1
preferred by each
node
R
0 2
October 16, 2002 Abhay K. Parekh: Topics in Routing 37
BGP Bouncing Problem (Labovitz et al)
Initially direct route
1
preferred by each
node
R has a fault so
R
sends withdrawal
messages to each
0 2
of its neighbors
October 16, 2002 Abhay K. Parekh: Topics in Routing 38
BGP Bouncing Problem (Labovitz et al)
Initially direct route
1
preferred by each
node
R has a fault so
R
sends withdrawal
messages to each
0 2
of its neighbors
October 16, 2002 Abhay K. Parekh: Topics in Routing 39
BGP Bouncing Problem (Labovitz et al)
Initially direct route
1
preferred by each
node
R has a fault so
R
sends withdrawal
messages to each
0 2
of its neighbors
October 16, 2002 Abhay K. Parekh: Topics in Routing 40
BGP Bouncing Problem (Labovitz et al)
Initially direct route
1
preferred by each
node
R has a fault so
R
sends withdrawal
messages to each
of its neighbors
0 2
Nodes choose
clockwise 2 hop
paths
October 16, 2002 Abhay K. Parekh: Topics in Routing 41
BGP Bouncing Problem (Labovitz et al)
Initially direct route
preferred by each node
1
R has a fault so sends
withdrawal messages to
each of its neighbors
R
Nodes choose clockwise
2 hop paths
Detecting loops they
0 2
choose anticlockwise 2
hop paths
October 16, 2002 Abhay K. Parekh: Topics in Routing 42
BGP Bouncing Problem (Labovitz et al)
Initially direct route
preferred by each node
1
R has a fault so sends
withdrawal messages to
each of its neighbors
R
Nodes choose clockwise
2 hop paths
Detecting loops they
0 2
choose anticlockwise 2
hop paths
October 16, 2002 Abhay K. Parekh: Topics in Routing 43
BGP Bouncing Problem (Labovitz et al)
Initially direct route
preferred by each node
1
R has a fault so sends
withdrawal messages to
each of its neighbors
Nodes choose clockwise
R
2 hop paths
Detecting loops they
choose anticlockwise 2
hop paths 0 2
And so on
N! steps, and even more
messages
October 16, 2002 Abhay K. Parekh: Topics in Routing 44
Other problems
Most router messages bogus
Routing table oscillations worse under path
vector when compared to distance vector
Timer interactions with router vendor
implementations create pathologies
Network Administrator configuration errors
can create catastrophic outages throughout
the network
October 16, 2002 Abhay K. Parekh: Topics in Routing 45

Precautions
Route Damping
Don t believe updates unless they make sense
Improves oscillations but adds to convergence time
More state
Keep history of received and sent messages
Improves situation but quite expensive
Router Configuration Restrictions
Save Path Vector Protocols (Griffin et al)
Human error potential is still considerable
Alternatives
Detour
Routing Feedback based approaches
Hard to change the installed base!
October 16, 2002 Abhay K. Parekh: Topics in Routing 46
Conclusions
Hierarchical Routing is a core component of most
large networks
Internet interconnection is complex and multi-
faceted
The ability for the network to expand its range of
services is tied to the modes of peering
Internet routing is becoming less hierarchical and
more driven by policy than performance
BGP stability is important and a work in progress
October 16, 2002 Abhay K. Parekh: Topics in Routing 47

References
Peering
Norton: Internet Service Providers and Peering
Eli Noam: Interconnecting the Network of Networks , MIT
Press, 2001.
BGP: Basics
BGP Tutorial by Nina Taft, SprintLabs
Halabi, Internet Routing Architectures , Cisco Press 1997
BGP: Instability
Labovitz et al (see readings)
Griffin et al (see readings)
October 16, 2002 Abhay K. Parekh: Topics in Routing 48