Pathlet Routing Brighten Godfrey Scott Shenker Ion Stoica - - PowerPoint PPT Presentation

▶

Mar 17, 2024 341 likes •558 views

Pathlet Routing Brighten Godfrey Scott Shenker Ion Stoica {pbg,shenker,istoica}@cs.berkeley.edu UC Berkeley Hotnets 2008 multipath internet routing good for everyone! reliability source observes directly, reacts quickly path quality

SLIDE 1

Pathlet Routing

Brighten Godfrey Scott Shenker Ion Stoica {pbg,shenker,istoica}@cs.berkeley.edu UC Berkeley Hotnets 2008

SLIDE 2

reliability path quality money source observes directly, reacts quickly source observes directly, knows what it wants network providers can sell new service good for everyone!

multipath internet routing

SLIDE 3

“The Route Not Taken” [Frost 1920] Two routes diverged in a network, and I -- I took the one less transited, And that has reduced latency by up to 41%.

good for everyone

SLIDE 4

why don’t we have it?

Even if everyone involved wants multipath, no way to do it in BGP! All paths blocked except one. Offer more? State explosion!

SLIDE 5

why don’t we have it?

AS-level source routing... ...gives network owners no control. a d c b

b,c,d

SLIDE 6

why don’t we have it?

great at blocking paths! bad at allowing them. path vector (BGP) AS-level source routing great at allowing paths! bad at blocking them. Highly constrained routing policies. flexible policy control many paths goal:

SLIDE 7

pathlet routing

source routing + fragments of paths (pathlets) flexible policy control many paths goal:

SLIDE 8

utline
the protocol
emulating other protocols
local transit (LT) policies

SLIDE 9

pathlet routing

example

a b d c

1. vnode: virtual node

within an AS

2. pathlet: sequence
f vnodes
3. announce pathlets
4. source lists

pathlets in packet

ne per AS
ne per link

gossip all known pathlets to neighbors

= AS level source routing

128.2.0.0/16

SLIDE 10

forwarding plane

pathlets tagged with

Forwarding ID sequence

packet contains list
f FIDs
forwarding table

maps FID to, e.g.,

utgoing interface

a b d c

128.2.0.0/16

1 2 3 1 1 2 1 2

1,2,2 2,2 2

SLIDE 11

utline
the protocol
emulating other protocols
local transit (LT) policies

SLIDE 12

emulating BGP

a b d c

128.2.0.0/16

1 1 7,1 1,7,1

1,7,1 7,1 1

(not discussed: how to verify route follows advertised policy)

SLIDE 13

MIRO [Xu, Rexford, SIGCOMM’06]
NIRA [Yang, Clark, Berger, ToN’07]

emulating other protocols

SLIDE 14

utline
the protocol
emulating other protocols
local transit (LT) policies

SLIDE 15

“local transit” policies

Each ingress --> egress pair is either allowed or disallowed. Subject to this, any path allowed!

SLIDE 16

LTP a common case?

capture a network’s direct costs
valley freeness is a LT policy, and the

common case in BGP export policies today

SLIDE 17

valley free routing as an LTP

peers providers customers “customers can route to anyone; anyone can route to customers”

SLIDE 18

scalability

BGP Pathlet routing, class-based LT policies forwarding table entries O(kn) O(d+k) control plane entries O(dkn) O((d+k)n) control plane messaging > O(Lkn/d) O(dn) n = # ASes d = mean # neighbors L = mean path len k = prefixes per AS

SLIDE 19

forwarding table entries

BGP pathlet routing, LT policies

ne per destination

(IP prefix)

ne for each pathlet

starting at the router current Internet (CAIDA/APNIC): 266,073 entries 2,317 entries, max 6 entries, mean

SLIDE 20

conclusion

pathlet routing: flexible policies --> multipath

with many choices, better scalability

can’t emulate everything, e.g. FBR [Zhu, Gritter,

Cheriton ‘03]

emulate others? path splicing [Motiwala, Elmore,

Feamster, Vempala 2008], Routing Deflections [Yang, Wetherall 2006]

challenge for all multipath protocols:

different payment for different paths?

thanks: fonts by tom7