Frame Relay

cs.vsb.cz

Frame Relay

Frame Relay

Petr Grygárek

rek

© 2005 Petr Grygarek, Advanced Computer Networks Technologies 1


Basic characteristics

• Used to implement WAN links over shared infrastructure

• infrastructure provided by independent operator

• private Frame-Relay infrastructure

• Layer 2 technology

• Assumes fast and reliable links between switching elements in FR cloud

• No hop-by-hop error-control and flow-control

• upper layers of user devices have to solve it by themselves

• Fast and efficient switching

• Independent of layer 3 protocol(s) used

• Virtual-circuit based

• implemented using fast frame (L2) switching (“relaying”)

• Fast VC creation/deletion (compared to leased line)

• QoS-enabled

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

2


Infrastructure and DTEs

Most-common DTEs

• Router

Frame Relay Access Device (FRAD)

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

3


Virtual circuits

• Set-up over FR infrastructure (cloud)

• Simulate “real” circuits with predefined parameters

• Permanent Virtual Circuit (PVC)

• requires (software based) preconfiguration by infrastructure provider

• used most often

• Switched Virtual Circuit (SVC)

• created on user-device (DTE) demand

• need for signalling standard and unique DTE addresses

• Q.933 – similar to ISDN Q.931

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

4


VC Principle

FR Switch

I 1

I 2

I 1

DLCI in

I 2

I in

FR Switch

110

DLCI out I out

----------------------------------------

110 I 2

17 I 3

17 I 3

110 I 2

I 3 17

DTE B

I 1

I 2

I 1

FR Switch

56 121

DTE A

I 3

DLCI in

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

I in

DLCI out I out

----------------------------------------

56 I 3

121 I 2

121 I 2

56 I 3

I 2

FR Switch

DLCI in

I in

DLCI out I out

----------------------------------------

121 I 1

110 I 2

110 I 2

121 I 1

5


Data Link Connection Identifier

(DLCI)

• Identifies individual virtual circuits on local loop

• and on inter-FR-switch links, if network cloud implemented

using FR switches

• 10-bit value

• Locally unique

• can be repeated on different links

• Assigned by FR cloud operator

• User DLCI: 16 to 1007

Frame Relay switches change DLCI along PVC path

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

6


Virtual circuit configuration

(FR switch-based infrastructure)

• In each FR switch along PVC path, switching

table entry must be entered


• Both PVC direction must be configured

• the same DLCIs used on each link in both directions

commonly

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

7


DLCI local significance

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

8


Virtual circuit advantages

• Requires only one physical access line regardless

to number of virtual circuits established

• doesn’t require multiple leased lines to customer site

• per PVC payment, but smaller fees compared to

separate leased lines

• Flexible – new circuits can be quickly added and

unnecessary ones deleted on as-needed basis

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

9


Switched virtual circuits (SVC)

• Q.933 signalling between user and FR switch

• subset of Q.931

• Sent over DLCI 0 (together with LMI messages)

• Q.922 provides reliable link for Q.933 messages transfer

• QoS parameters appointed during call setup

• Connection establishment:

• SETUP, PROGRESS, ALERTING, CALL PROCEEDING,

CONNECT, CONNECT ACK

• Connection Termination:

• DISCONNECT, RELEASE, RELEASE COMPLETE

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

10


Local access loop

•Subscriber connected by single leased line (or ISDN)

•Typical access line rates 56kbps to (approx) 54 Mbps

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

11


Frame Relay standard

Frame-relay defines es local access loop, not

implementation of network cloud

(infrastructure)

• may be built upon frame-relay switches, ATM,

connectionless IP or other technology

• invisible to Frame Relay user

• Infrastructure must provide committed

parameters to provisioned VCs

• CAR, Bc, Be, …

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

12


Frame Relay Standardization Bodies

1. Gang of Four:

1.

• DEC, Cisco, Northern Telecom, StrataCom …

2. www.frforum.com

2.

• Cooperates with ITU-T

• FRF.x standards

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

13


Frame Relay Interfaces

• User-to-Network Interface (UNI)

• between customer and FR operator

• Network-to-Network Interface (UNI)

• between two FR operators

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

14


User-to-Network Interface

(UNI)

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

15


DTE to DCE protocol (Q.922)

Frame delimited by flags (bit stuffing used)

Frame based on HDLC

• Variable frame length (max 8/16kB)

• Multiple frame formats (historical reasons)

• IETF (RFC1490)

• (older) CISCO frame: 4B header

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

16


Frame header fields (1)

• DLCI – Data Link Connection Identifier

• C/R (Command/Response) bit

• currently unused (maintained for compatibility with

HDLC)

• EA (Extended Address)

• if set to 1, current byte is the last addressing byte

(otherwise another byte follows)

• Provides capability to extend DLCI in the future

• Current implementations all use 2B address field

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

17


Frame header fields (2)

Fields for congestion avoidance and resolution

• FECN – Forward Explicit Congestion

Notification

• BECN – Backward Explicit Congestion

Notification

• DE – Discard Eligibility

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

18


Multiprotocol encapsulation

Frame header does not carry layer 3 protocol ID

• FRF.3/RFC 1490 - Multiprotocol Interconnect over Frame Relay

• L3 protocol given by NLPID

NLPID (Network Layer Protocol ID)

• 2 bytes just behind frame header

• NLPID values defined in ISO/IEC TR 9577

• NLPID may indicate usage of SNAP header

• Earlier proprietary L3 protocol ID identification

Can vary on different VCs

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

19


Network to Network Interface

(NNI)

• Connects two different FR clouds

• Supports multi-network PVCs

• Consists of PVC segments (over individual FR

clouds)

• DLCI assignment and QoS parameters setting

coordination needed on NNI

• Bridges signalling protocols variations

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

20


Local Management Interface

(LMI)

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

21


Local Management Interface (LMI)

• Operates between DTE and FR access switch

• FR DTE, FR DCE

• Multiple variations (historical reasons)

• ANSI T1.617 (=Annex D), q933a(=Annex A), Gang

of Four

• DLCI 0 / 1023

• VC status messages

• DTE-DCE line keepalives

• Obtaining list of VCs from FR switch

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

22


LMI frame

• Carried in normal FR frame

• reserved protocol number (0x09)

• Reserved DLCI

• Call reference – unused

• Message Type

• Status-inquiry

• Status-message (keepalive), Full Status Message (PVC list)

• differ in IEs included

• Variable number of information elements (IE)

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

23


Frame Relay Congestion

Management

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

24


Congestion avoidance and recovery

• Congestion avoidance

• ability to inform stations about emerging congestion

• Congestion recovery

• actions taken to recover from congestion and return

to normal operating state

• frame discard

Separate for every VC and direction

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

25


Congestion avoidance

implementation

• Backward

Explicit

Congestion

Notification (BECN)

• Set by intermediate switching elements to ask source to slow

down

• Cisco: source slows down by 25 % (if not under configured minimal

imal

CIR)

• Forward Explicit

Congestion

Notification (FECN(

FECN)

• Set by intermediate switching elements to inform destination

it should ask source to slow down

• Destination utilizes upper layer protocol to accomplish that

• (e.g. TCP starts to delay ACKs)

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

26


Congestion recovery

implementation

• Discard Eligibility (DE) bit

• Set by access FR switch for nonconforming frames

• If congestion occurs on some link, frames with DE

set should be dropped first

• May be set by source DTE to give priorities to

traffic entering FR cloud

• If congestion persists even when DE frames

discarded, network have to start discarding of

non-DE frames

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

27


Frame Relay QoS parameters,

their measurement and enforcement

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

28


FR virtual circuit QoS parameters

Parameters take into account bursty nature of traffic

• Bc – committed burst

• Maximum amount of data the network commits to deliver over a VC during

appointed interval Tc

• Committed information rate (CIR)

• estimate of normal traffic during a busy period

• CIR=Bc/Tc

• Sum of CIRs of all VCs commonly greater than physical access line speed

• Be - excess burst

• Maximum amount of data by which a user can exceed Bc during an interval Tc

• Delivered at lower probability than the data within Bc

• Excess Information Rate (EIR)

• EIR=Be/Tc

Frame above CIR+EIR (maximum rate) are discarded

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

29


Service Level Agreement

Access rate = physical access line bitrate

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

30


Traffic policing and traffic shaping

• Traffic policing – done at provider’s network

edge to limit traffic coming from customer

• Traffic exceeding committed information rate is

either discarded or marked as eligible for later

discard

• Traffic shaping – done at customer boundary

router to shape traffic to comply with rate

provider agrees to transfer

• uses memory buffers to shape bursts

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

31


Traffic policing

• Runs on FR access switch

• Determines whether to pass incoming frame to

network cloud, mark it as discard eligible or

discard it

• No further traffic flow measurements are taken

inside network cloud

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

32


Traffic shaping /rate limiting

• Applied at customer boundary router

• Mechanism to smooth (bursty) traffic flow to

meet service provider requirements (CIR,Bc,Be)

Frames exceeding agreement are buffered and

sent later

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

33


Traffic policing/shaping

implementation algorithms

• Leaky bucket

• imposes a hard limit on the data transmission rate

• Token bucket

• allows a certain amount of burstiness while imposing

a limit on the average data transmission rate

• sometimes considered an improvement of Leaky

bucket

Leaky bucket and token bucket algorithms are

often mistakenly lumped together

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

34


Leaky bucket Algorithm

Provides a mechanism by which bursty traffic can be shaped into

steady stream of traffic to the network

• Arriving frames are placed in a bucket with a hole in the bottom

• The bucket can queue at most b bytes. If a frame arrives when

the bucket is full, the frame is discarded.

Frames s drain through the hole in the bucket, into the network, at

a constant rate of r bytes per second, thus smoothing traffic

bursts.

• In practise, we need to sent not exactly r bytes, but whole frame

Ineffective, do not allow to pass traffic bursts even if network

bandwidth is available

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

35


Token Bucket Algorithm

• Once per time interval (Tc), tokens are added to the

bucket

• Number corresponds to CIR in Bytes/second

• Bucket has limited capacity (Bc+Be), overleaking tokens are

discarded

• If frame arrives, token bucket is checked whether there

is at least as many tokens as data byte count in

incoming frame

• If there is enough tokens, frame is passed and appropriate

number of tokens removed from the bucket

• Otherwise, the frame is stored into waiting queue (buffered) /

discarded

• Be e allows to “save” some tokens from previous Tc

interval(s) to the current one

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

36


Token Bucket Implementation

• FR access switch maintains a cumulative amount

of data sent by user (counter C)

• C decremented by Bc every Tc interval

• C is always greater than 0

• When frame arrives

• If C Bc+Be: Discard frame

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

37


Typical rate limiting usage

Often physical local loop rate at central site

outperforms physical local loop rate at branch

site

• Because we want to communicate with multiple

branch sites

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

38


TCP/IP over Frame Relay

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

39


Scenerio

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

40


Typical Frame Relay usage

topologies

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

41


Typical usage

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

42


Relationship between IP subnets and

FR virtual circuits

• One physical router interface may correspond to

one VC

• One physical router interface may accommodate

multiple VCs (more(

obvious)

• logical subinterface assigned to every VC

• bound to right VC by DLCI configuration

• behaves like physical interface from routing

protocol’s point of view

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

43


Subinterfaces

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

44


Frame Relay (sub)interface types

• Point-to-point

• separate IP subnet for every subinterface

• Point-to-multipoint

• common IP subnet on all VCs

• similar to broadcast network, but without broadcast

capability

• router interface “routes” between connected VCs

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

45


IP to DLCI mapping

• Point-to-point interface

• Single IP address and DLCI configured

• Point-to-multipoint interface

• IP address of local side of (all) VCs configured

• Static maps configured

• for every possible peer IP address we configure DLCI of

VC connected to that peer

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

46


Inverse ARP (INARP)

INARP allows automatic discovery of peer IP address to

DLCI mapping

• If no DLCI or static map is configured on

(sub)interface, router sends INARP FR Frame to every

known VC

• List of available DLCIs can be obtained from FR switch

using LMI

• Peer responds with L3 address configured on it’s

interface

• DLCI-to-peer-address mapping dynamic entry can be

created

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

47


Frame-relay routing problems

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

48


DV routing updates problems

• Distance vector routing protocols use broadcast

to send routing updates

• Cisco implementation: VCs do not pass

broadcasts if not explicitly configured to do so

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

49


Split Horizon problem

Split-horizon rule must be abandoned on physical

interface accommodating multiple VCs

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

50


OSPF

Frame-relay cloud represented with NBMA network

node type in OSPF topology graph

• Non-Broadcast Multi Access = multiple neighbors accessible

via single physical interface, but no single broadcast frame

can be sent to all of them

• Be careful when configuring DR priorities on NBMA

network with other than full-mesh topology

• DR should be directly connected to all other routers using

VC

• Different OSPF timers used on NBMA networks

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

51


FR Additional features

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

52


FR multicast support

• All multicast traffic of a multicast group sent to

entity called multicast server, which forwards

data to multicast group members

• DLCI 1019-1022 reserved for that purpose

• One-way and two-way multicast

• Additional LMI messages defined to notify user

about addition, deletion and presence of

multicast group

Not supported by most operators

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

53


Interesting FR references

• http://www.cisco

cisco.com/univercd/cc/td/doc/cisintwk/

ito_doc

doc/frame.htm

• Overview and basic concepts

• http://www.protocols.com/pbook/frame.htm

• Detailed overview of FR standards

• http://www.cse.wustl.edu/~jain/cis788-95/ftp/frame_relay

• Advanced FR features

• http://qbone.internet2.edu/bb/Bucket.doc

• http://en.wikipedia.org/wiki/Token_bucket

• Token Bucket/Leaky Bucket algorithm

© 2005 Petr Grygarek, Advanced Computer Networks Technologies

54

More magazines by this user
Similar magazines