C ONTENTS February 2009 Introduction Ethernet and Spanning Tree - - PowerPoint PPT Presentation

c ontents
SMART_READER_LITE
LIVE PREVIEW

C ONTENTS February 2009 Introduction Ethernet and Spanning Tree - - PowerPoint PPT Presentation

February 2009 RB RIDGES L2 Forwarding with Link State Routing L AYER 2 F ORWARDING B ASED ON L INK S TATE R OUTING Donald E. Eastlake 3 rd 1 donald.eastlake@stellarswitches.com C ONTENTS February 2009 Introduction Ethernet and Spanning


slide-1
SLIDE 1

RBRIDGES

LAYER 2 FORWARDING BASED ON LINK STATE ROUTING

Donald E. Eastlake 3rd

donald.eastlake@stellarswitches.com

February 2009

1

L2 Forwarding with Link State Routing

slide-2
SLIDE 2

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

2

L2 Forwarding with Link State Routing

slide-3
SLIDE 3

DEFINITIONS

 RBridge – Routing Bridge  A device implementing the TRILL protocol, which

performs Layer 2 bridging with link state routing.

 RBridge Campus –  A network of RBridges, links, and possibly

intervening bridges bounded by end stations.

 TRILL –

TRansparent Interconnection of Lots of Links

 A standard being specified by the IETF (Internet

Engineering Task Force) TRILL Working Group co- chaired by

 Donald E. Eastlake 3rd, Stellar Switches  Erik Nordmark, Sun Microsystems

February 2009

3

L2 Forwarding with Link State Routing

slide-4
SLIDE 4

WHY/WHO RBRIDGES/TRILL?

 Why do RBridges/TRILL?  Provide optimum point-to-point forwarding with zero

configuration.

 Support multi-pathing of both unicast and multi-

destination traffic.

 Who invented RBridges/TRILL?  Radia Perlman of Sun Microsystems, also the

inventor of the Spanning Tree Protocol.

February 2009

4

L2 Forwarding with Link State Routing

slide-5
SLIDE 5

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

5

L2 Forwarding with Link State Routing

slide-6
SLIDE 6

Ethernet

 Invented in the 1970s by Bob Metcalfe At Xerox  Carrier Sense Multiple Access Collision Detect

(CSMA/CD)

 DIX (Digital, Intel, Xerox) agree around 1980  IEEE Standardization started around 1983,

completed in 1985

 Ever increasing speed for wired/optical-fiber:  <10Mbps

10Mbps

 100Mbps

1Gbps

 10Gbps  Under development: 40Gbps, 100Gbps

February 2009

6

L2 Forwarding with Link State Routing

slide-7
SLIDE 7

Ethernet Local Area Network (LAN) Evolution

 Multi-access media  Repeaters  Hubs – full duplex  Bridges, learning

February 2009

7

L2 Forwarding with Link State Routing

slide-8
SLIDE 8

Ethernet Local Area Network (LAN) Evolution

 Hubs – full duplex  Bridges, learning

February 2009

8

L2 Forwarding with Link State Routing

slide-9
SLIDE 9

Ethernet Local Area Network (LAN) Evolution

 Bridges  Spanning Tree Protocol invented by Radia in 1985  Address Learning and Forgetting

February 2009

9

L2 Forwarding with Link State Routing

slide-10
SLIDE 10

Algorhyme

 I think that I shall never see

a graph more lovely than a tree.

 A tree whose crucial property

is loop-free connectivity.

 A tree that must be sure to span

so packets can reach every LAN.

 First, the root must be selected.

By ID, it is elected.

 Least-cost paths from root are traced.

In the tree, these paths are placed.

 A mesh is made by folks like me,

then bridges find a spanning tree.

 Radia Perlman

February 2009

10

L2 Forwarding with Link State Routing

slide-11
SLIDE 11

Spanning Tree Difficulties

 The Spanning Tree Protocol makes a general

mesh of connected bridges into a tree by disabling

  • ports. This means that

 traffic is concentrated on the remaining links,

increasing congestion, and

 traffic is not pair-wise shortest path but must follow

whatever path is left after spanning tree blocks redundant paths.

February 2009

11

L2 Forwarding with Link State Routing

slide-12
SLIDE 12

Spanning Tree Difficulties

 There is no hop count in Ethernet, which makes

temporary loops more dangerous. Loops can appear with spanning tree due to

 sufficient dropped spanning tree messages, or  the appearance of new connectivity without physical

indication.

 Failover minimum time limitations for some

failures.

 Connectivity changes can cause VLANs to

partition.

February 2009

12

L2 Forwarding with Link State Routing

slide-13
SLIDE 13

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

13

L2 Forwarding with Link State Routing

slide-14
SLIDE 14

OPTIMUM POINT-TO-POINT FORWARDING

February 2009

14

=
end
sta)on
 
B2
 B3
 B1


A three bridge network

L2 Forwarding with Link State Routing

slide-15
SLIDE 15

OPTIMUM POINT-TO-POINT FORWARDING

February 2009

15

Spanning tree eliminates loops by disabling ports

=
end
sta)on
 
B2
 B3
 B1


L2 Forwarding with Link State Routing

slide-16
SLIDE 16

OPTIMUM POINT-TO-POINT FORWARDING

February 2009

16

RB2
 =
end
sta)on
 RB3
 RB1


A three RBridge network: better performance using all facilities

L2 Forwarding with Link State Routing

slide-17
SLIDE 17

MULTI-PATHING

February 2009

17

B2
 =
end
sta)on
 B4
 B3
 B1


Bridges limit traffic to one path

L2 Forwarding with Link State Routing

slide-18
SLIDE 18

MULTI-PATHING

February 2009

18

RB2
 =
end
sta)on
 RB4
 RB3
 RB1


RBridges support multi-path for higher throughput

L2 Forwarding with Link State Routing

slide-19
SLIDE 19

Other RBridge Features

 Compatible with classic bridges. Can be

incrementally deployed into a bridged LAN.

 Forwarding tables at transit RBridges scale with

the number of RBridges, not the number of end

  • stations. Transit RBridges do not learn end

station addresses.

 A flexible options feature. RBridges know what

  • ptions other RBridges support.

 Globally optimized distribution of IP derived

multicast.

February 2009

19

L2 Forwarding with Link State Routing

slide-20
SLIDE 20

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

20

L2 Forwarding with Link State Routing

slide-21
SLIDE 21

THE TRILL ENCAPSULATION AND HEADER

 Frames sent between RBridges are encapsulated

inside a local link header, addressed from the local source RBridge to the local destination RBridge, and a TRILL header.

February 2009

21

RBridge One RBridge Two Ethernet Cloud

L2 Forwarding with Link State Routing

slide-22
SLIDE 22

THE TRILL ENCAPSULATION AND HEADER

 Some reasons for encapsulation:  Provides a hop count to mitigate loop issues  To hide the original source address to avoid confusing

any bridges present as might happen if multi-pathing were in use

 To direct unicast frames toward the egress RBridge

so that forwarding tables in transit RBridges need

  • nly be sized with the number of RBridges in the

campus, not the number of end stations

 To provide a separate VLAN tag for forwarding

traffic between RBridges, independent of the original VLAN of the frame

February 2009

22

L2 Forwarding with Link State Routing

slide-23
SLIDE 23

THE TRILL ENCAPSULATION AND HEADER

 Assuming the link is Ethernet (IEEE 802.3) the

encapsulation looks like:

1.

Outer Ethernet Header

Source RBridge One, Destination RBridge Two

2.

(Outer VLAN Tag)

3.

TRILL Header

4.

Inner Ethernet Header

Original Source and Destination Addresses

5.

Inner VLAN Tag

6.

Original Payload

7.

Frame Check Sequence (FCS)

February 2009

23

L2 Forwarding with Link State Routing

slide-24
SLIDE 24

THE TRILL ENCAPSULATION AND HEADER

 TRILL Header – 64 bits  Nicknames – auto-configured 16-bit campus local

names for RBridges

 V = Version (2 bits)  R = Reserved (2 bits)  M = Multi-Destination (1 bit)  OpLng = Length of TRILL Options  Hop = Hop Limit (6 bits)

February 2009

24

TRILL Ethertype Egress RBridge Nickname Hop OpLng V

M

R Ingress RBridge Nickname

L2 Forwarding with Link State Routing

slide-25
SLIDE 25

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

25

L2 Forwarding with Link State Routing

slide-26
SLIDE 26

ARE RBRIDGES BRIDGES OR ROUTERS?

 They are obviously Bridges because  RBridges deliver unmodified frames from the source

end station to the destination end station

 RBridges can operate with zero configuration and

auto-configure themselves

 RBridges provide the restriction of frames to VLANs

as IEEE 802.1Q bridges do

 RBridges can support frame priorities as IEEE

802.1Q bridges do

 RBridges, by default, learn MAC addresses from the

data frames they receive

February 2009

26

L2 Forwarding with Link State Routing

slide-27
SLIDE 27

ARE RBRIDGES BRIDGES OR ROUTERS?

 They are obviously Routers because  RBridges decrement a hop count in TRILL frames on

each hop

 RBridges swap the outer addresses on each RBridge

hop from the ingress RBridge to the egress RBridge

 RBridges use a routing protocol rather than the

spanning tree protocol

 RBridges optionally learn MAC addresses by

distribution through the control messages

 RBridges normally act based on IP multicast control

messages (IGMP, MLD, and MRD) and restrict the distribution of IP derived multicast frames

February 2009

27

L2 Forwarding with Link State Routing

slide-28
SLIDE 28

ARE RBRIDGES BRIDGES OR ROUTERS?

 Really, they are a new species, between IEEE

802.1 bridges and routers:

February 2009

28

Routers

(plus servers and other end stations)

RBridges Bridges Hubs/Repeaters

L2 Forwarding with Link State Routing

Bridged LAN RBridge Campus

slide-29
SLIDE 29

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

29

L2 Forwarding with Link State Routing

slide-30
SLIDE 30

WHY IS-IS FOR TRILL?

 The IS-IS (Intermediate System to Intermediate

System) link state routing protocol was chosen for TRILL over OSPF (Open Shortest Path First), the only other candidate, for the following reasons:

 IS-IS runs directly at Layer 2. Thus no IP addresses

are needed, as they are for OSPF, and IS-IS can run with zero configuration.

 IS-IS uses a TLV (type, length, value) encoding which

makes it easy to define and carry new types of data.

 (IS-IS is the international standard which grew

  • ut of DECnet Phase V, in which Radia Perlman

was heavily involved.)

February 2009

30

L2 Forwarding with Link State Routing

slide-31
SLIDE 31

HOW RBRIDGES WORK

 RBridges find each other by exchanging TRILL

IS-IS Hello frames

 Like all TRILL IS-IS frames, TRILL Hellos are sent

to the All-IS-IS-RBridges multicast address. They are transparently forwarded by bridges, dropped by end stations including routers, and are processed (but not forwarded) by RBridge ports.

 The Hellos establish connectivity on each port.  Using the information exchanged in the Hellos, the

RBridges on each link elect the Designated RBridge for that link

February 2009

31

L2 Forwarding with Link State Routing

slide-32
SLIDE 32

HOW RBRIDGES WORK

 The Designated RBridge specifies the Appointed

Forwarder for each VLAN on the link (which may be itself) and the Designated VLAN for inter- RBridge communication.

 The Appointed Forwarder for VLAN-x on a link

handles all native frames to/from that link in that VLAN.

 It encapsulates native frames from the link into a

TRILL data frame, the ingress RBridge function.

 It decapsulates native frames destined for the link

from TRILL data frames. This is the egress RBridge function.

February 2009

32

L2 Forwarding with Link State Routing

slide-33
SLIDE 33

HOW RBRIDGES WORK

 RBridges use the IS-IS reliable flooding protocol

so that each RBridge has a copy of the global “link state” database.

 The RBridge link state includes information beyond

connectivity and link cost. Information such as VLAN connectivity, multicast listeners and multicast router attachment, claimed nickname, options supported, and the like.

 The database is sufficient for each RBridge to

independently and without further messages calculate optimal point-to-point paths for known unicast frames and the same distribution trees for multi-destination frames.

February 2009

33

L2 Forwarding with Link State Routing

slide-34
SLIDE 34

HOW RBRIDGES WORK

 TRILL data frames with  known unicast ultimate destinations are forwarded

RBridge hop by RBridge hop toward the egress RBridge.

 multi-destination frames (broadcast, multicast, and

unknown destination unicast) are forwarded on a tree rooted at an RBridge selected by the ingress RBridge.

February 2009

34

L2 Forwarding with Link State Routing

slide-35
SLIDE 35

Algorhyme V2

 I hope that we shall one day see

A graph more lovely than a tree.

 A graph to boost efficiency

While still configuration-free.

 A network where RBridges can

Route packets to their target LAN.

 The paths they find, to our elation,

Are least cost paths to destination.

 With packet hop counts we now see,

The network need not be loop-free.

 RBridges work transparently.

Without a common spanning tree.

 Ray Perlner

February 2009

35

L2 Forwarding with Link State Routing

slide-36
SLIDE 36

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

36

L2 Forwarding with Link State Routing

slide-37
SLIDE 37

STRUCTURE OF AN RBRIDGE

February 2009

37

Central Processing IS-IS, Mgmt., Etc. Switching fabric Port Logic Port Logic Port Logic Port Logic Links to other devices. Could be 802.3 (Ethernet), 802.11 (Wi-Fi), PPP, …

L2 Forwarding with Link State Routing

slide-38
SLIDE 38

STRUCTURE OF AN RBRIDGE PORT

February 2009

38

Assumes an Ethernet (802.3) link. Information Frame & Info 802.3 Physical Interface 802.3 Link 802.1/802.3 Low Level Control Frame Processing, Port/Link Control Logic 802.1Q Port VLAN Processing RBridge: High Level Control Frame Processing (BPDU, VRP) ISS EISS RBridge: Higher Level Processing (see next slide)

L2 Forwarding with Link State Routing

slide-39
SLIDE 39

STRUCTURE OF AN RBRIDGE PORT

February 2009

39

Information Frame & Info Lower Level Processing (see previous slide) RBridge: Inter-port Forwarding, IS-IS, Management, Etc. Appointed Forwarder and Inhibition Logic Native frames Encapsulation / Decapsulation Processing TRILL IS-IS Hello frames TRILL IS-IS Hello Processing TRILL data frames TRILL data and

  • ther TRILL

IS-IS frames

L2 Forwarding with Link State Routing

slide-40
SLIDE 40

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

40

L2 Forwarding with Link State Routing

slide-41
SLIDE 41

ADDRESS LEARNING

 From Local Frames  { VLAN, Source Address, Port }  From Decapsulated Frames  { Inner VLAN, Inner Source Address,

Ingress RBridge }

 The remote RBridge is learned as the proper egress RBridge

for frames sent to the remote address and VLAN

 Via Optional End Station Address Distribution

Information protocol

 { VLAN, Address, RBridge nickname }

February 2009

41

L2 Forwarding with Link State Routing

slide-42
SLIDE 42

WHAT ABOUT RE-ORDERING?

 RBridges are required to maintain frame

  • rdering internally, modulo frame priority.

 When multi-pathing is used, all frames for an

  • rder-dependent flow must be sent on the same

path if unicast or the same distribution tree if multi-destination.

 Re-ordering can occur briefly when a destination

address transitions between being known and unknown or a topology change occurs.

February 2009

42

L2 Forwarding with Link State Routing

slide-43
SLIDE 43

WHAT ABOUT LOOPS?

 TRILL Data Frame Loops:  Known unicast frames have a hop count and are

always unicast to the next hop RBridge.

 Multi-destination frames must be received on a port

which is part of their distribution tree, the ingress RBridge nickname must pass a Reverse Path Forwarding Check, and they have a hop count.

 Hybrid TRILL Data / Native Frame Loops:  TRILL takes great care to assure that there are

almost never two uninhibited appointed forwarders

  • n the same link for the same VLAN.

 Pure Native Frame Loops: Not TRILL’s problem.

February 2009

43

L2 Forwarding with Link State Routing

slide-44
SLIDE 44

CONTENTS

 Introduction  Ethernet and Spanning Tree  RBridge Features  TRILL Encapsulation  Are RBridges Bridges or Routers?  How RBridges Work  Structure of an RBridge  Some Additional Details  References

February 2009

44

L2 Forwarding with Link State Routing

slide-45
SLIDE 45

REFERENCES

 Specification Draft:

“Rbridges: Base Protocol Specification”

 http://tools.ietf.org/html/draft-ietf-trill-rbridge-

protocol-11

 Original Paper by Radia Perlman:

“Rbridges: Transparent Routing”

  • http://www.postel.org/rbridge/infocom04-paper.pdf

 Current TRILL WG Charter  http://www.ietf.org/html.charters/trill-charter.html  “TRILL: Problem and Applicability Statement”  http://www.ietf.org/internet-drafts/draft-ietf-trill-

prob-05.txt

February 2009

45

L2 Forwarding with Link State Routing

slide-46
SLIDE 46

END

Donald E. Eastlake 3rd

donald.eastlake@stellarswitches.com

February 2009

46

L2 Forwarding with Link State Routing