Tài liệu Wide Area Networks doc

www.cisco.comKeep All Graphics Inside This Box econ_0481_09_010.ppt Policing Policing is the QoS component that limits traffic flow to a configured bit rate: • With limited bursting cap

Trang 1

Wide Area Networks

Trang 2

Keep All Graphics Inside This Box

econ_0481_09_010.ppt

Section I Policy / Shaping

Trang 3

The purpose of the lesson is to quickly survey the new policing and traffic shaping

features in Cisco IOS Release 12.1, and to describe the problems they solve

Trang 4

econ_0481_09_010.ppt

Central Site

Frame Relay, ATM

128kbps 256kbps 512kbps 768kbps T1

Result:

Buffering = Delay or Dropped PacketsCustomer Problems to Solve

• Central to Remote Site Speed Mismatch

• Remote to Central Site Over-subscription

• Control use of shared LAN, WAN, MAN media

–Multi-Dwelling Unit (MDU)

The slide shows a Frame Relay or ATM network Pay close attention to the speeds

of the access lines to the remote sites on the left Suppose each site has a

Committed Information Rate (CIR) close to the access speed with bursting up to

the access bandwidth

• What happens at the central site if the bottom two sites burst at the same

time?

• What happens at the central site if a server rapidly transmits data for the top

left remote site?

• What happens if the bottom two left sites try to send a large amount of data

to the top left site?

In this section, some of the QoS techniques that help resolves issues such as

theseare examined

Trang 5

econ_0481_09_010.ppt

Internet

Policing and traffic shaping occur within the network to provide congestion management and control bursts

Network ManagementPolicing and Shaping

In this module section, policing and traffic shaping are discussed Both of these

traffic engineering methods occur within the network as indicated by the heavy

ellipse in the slide They use the already marked Type of Service (ToS) or

Differentiated Services Code Point (DSCP) bits discussed in the previous module

With policing the rate at which traffic can flow is capped This is usually done

inbound to control how fast someone sends data

With shaping, smooth out bursts for a steadier flow of data Reduced burstiness

helps reduce congestion in a network core

Trang 6

econ_0481_09_010.ppt

Policing

Policing is the QoS component that limits traffic flow to a configured bit rate:

• With limited bursting capability

burst rate are dropped or have their precedence altered

A policer typically drops traffic

For example, CARs rate-limiting policer will either drop the packet or rewrite its

IP Precedence, resetting the packet header's ToS bits

Policing is also available through the MQC

Trang 7

econ_0481_09_010.ppt

Shaping

Shaping is the QoS feature component that regulates traffic flow to an average

or peak bit rate:

• With bursting capability

queued

A shaper typically delays excess traffic using a buffer, or queuing mechanism, to

hold packets and shape the flow when the data rate of the source is higher than

expected

For example, Generic Traffic Shaping (GTS) uses a weighted fair queue to delay

packets in order to shape the flow

Depending on how it is configured, Frame Relay Traffic Shaping (FRTS) uses

either a Priority Queue (PQ), a Custom Queue (CQ), or a first- in, first-out (FIFO)

queue for the same sort of purpose

Trang 8

econ_0481_09_010.ppt

Time

Time Traffic Rate

Traffic Policing Versus Shaping

Policer

Causes TCP resends Oscillation of TCP windows

Policer can be marker also (CAR)

Policer on input interface only

Shaper

Can adapt to network congestion (FR BECN, FECN)

This diagram shows the effects of traffic shaping

Both policing and shaping ensure that traffic does not exceed a (contracted) bandwidth

limit Policing and Shaping both limit bandwidth but with different traffic impact:

• Policing drops more often, more resends

• Shaping adds variable delay

Traffic shaping smoothes traffic by storing traffic above the configured rate in a queue

When a packet arrives at the interface for transmission, the following happens:

• If the queue is empty, the arriving packet is processed by the traffic shaper:

– If possible, the traffic shaper sends the packet

– Otherwise, the packet is placed in the queue

• If the queue is not empty, the packet is placed in the queue

When there are packets in the queue, the traffic shaper removes the number of packets it can send from the queue every time interval

Additional details on policing and shaping can be found at:

http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/12cgcr/qos_c/qcpart4/qcpolts.htm

Trang 9

Trang 10

econ_0481_09_010.ppt

Committed Access Rate (CAR)

CAR performs three functions:

and QoS group setting

limiting (policing)

CARs rate- limiting feature manages a network's access bandwidth policy by ensuring

that traffic falling within specified rate parameters is sent, while dropping packets that

exceed the acceptable amount of traffic or sending them with a different priority CARsexceed action is to drop packets

The rate- limiting function of CAR does the following:

• Allows the control the maximum rate of traffic transmitted or received on an

interface

• Gives the ability to define Layer 3 aggregate or granular incoming or outgoing

(ingress or egress) bandwidth rate limits and to specify traffic-handling policies

when the traffic either conforms to or exceeds the specified rate limits

• Uses aggregate bandwidth rate limits to match all of the packets on an interface or

subinterface

• Uses granular bandwidth rate limits to match a particular type of traffic based on

precedence, MAC address, or other parameters

CAR is often configured on interfaces at the edge of a network to limit traffic into or

out of the network

VIP-distributed CAR is a version of CAR that runs on the Versatile Interface Processor (VIP) It is supported on the Cisco 7500 routers with a VIP2-40 or greater interface

processor

Distributed Cisco Express Forwarding (dCEF) switching must be enabled on any

interface that uses VIP-Distributed CAR, even when only output CAR is configured

Trang 11

econ_0481_09_010.ppt

VoIP HTTP FTP

FTP HTTP

VoIP Gold Class Silver Class Bronze Class Separate “Conform” and

“Exceed” Actions

Policing Engine

VoIP HTTP FTP

CAR Marking and Policing

flexible rules

– IP Precedence / IP access list / incoming interface / MAC address

Once a packet has been measured as conforming to or exceeding a particular rate

limit, the router performs one of the following actions on the packet:

• Transmit—The packet is sent

• Drop—The packet is discarded

• Set precedence (or perhaps DSCP bits) and transmit—The IP Precedence

(ToS) bits in the packet header are rewritten The packet is then sent Use this

action to either color (set precedence) or recolor (modify existing packet

precedence) the packet

• Continue—The packet is evaluated using the next rate policy in a chain of rate limits If there is not another rate policy, the packet is sent

Trang 12

classification, for example,

drop low priority via WRED

if burst limit exceeded

1) Packet marking through IP Precedence and QoS group settings.

Based on ACL or inbound interface

2) Apply rate limiting to matching traffic pattern, for example, 25Kbps of traffic to “Bronze”

San Jose

Ottawa

CAR Bandwidth Management

Today, all the packets on the network look the same, and are thus handled the same, with each packet getting best-effort service CAR provides the capability to allow the service provider or enterprise to specify a policy which determines which packets should be

assigned to which traffic class The IP header already provides a mechanism to do this,

namely the three precedence bits in the ToS field in the IP header

CAR can set policies based on information in the IP or TCP header such as IP address,

application port, physical port or subinterface, and IP protocol to decide how the

precedence bits should be marked or “colored.” Once marked, appropriate treatment can

be given in the backbone to ensure that premium packets get premium service in terms of bandwidth allocation, delay control, and so on

CAR can also be used to police precedence bits set externally to the network either by the customer or by a downstream service provider Thus the network can decide to either

accept or override external decisions

CARs purpose is to identify packets of interest for packet classification or rate limiting or both, matching a specification such as:

1) All traffic

2) IP Precedence

3) MAC address

4) IP access list, standard and extended (slower)

See the following URL for additional information:

http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/12cgcr/qos_c/qcpart1/qccar.htm

Trang 13

econ_0481_09_010.ppt

CAR Action Policies

Precedence bits and transmit)

Precedence bits and go to the next rate-limit or police statement in the list)

In Release 11.1 CC the CAR rate limit list is not bounded as to length

Each CAR rate limit statement is checked sequentially for a match When a match

is found the token bucket, if there is one, is evaluated

If the action is a “continue” action, the policer will go to the next rate- limit on the

list to find a subsequent match If a match is found, the traffic is subjected to the

next applicable rate- limit

If an end of rate- limit list is encountered without finding a match or “continue”

action, the default behavior is to transmit

Trang 14

The slide lists the actions available with MQC police They are similar to, but

different than, the CAR action options

For additional information see:

http://www.cisco.com/univercd/cc/td/doc/product/aggr/10000/10ksw/qosos.htm#4

7183

Trang 15

The slide lists the actions available with MQC police They are similar to, but

different than, the CAR action options

http://www.cisco.com/univercd/cc/td/doc/product/aggr/10000/10ksw/qosos.htm#4

7183

Trang 16

econ_0481_09_010.ppt

Additional Policy Map Policing

and Shaping Options

Additional Policy Map Policing

and Shaping Options

Policing policy-map options:

Distributed Traffic Shaping (DTS) policy-map options:

The commands shown are some of the other options to use in the MQC policy

map They are listed here so all options can be referenced back to this location in

the module section

DTS commands will be covered in more detail later in this module To turn on

DTS, enter any of the shape commands

Trang 17

There are other options in the MQC policy map The options shown in the slide

invoke queuing methods (covered in more detail in the Queuing and Scheduling

module)

For example, to turn on WRED within a policy map, use any of the

random-detect commands To reserve a minimum bandwidth with CBWFQ, enter

the bandwidth command

Trang 18

econ_0481_09_010.ppt

Topics

Policing Traffic shaping

Trang 19

econ_0481_09_010.ppt

• LAN traffic tends to be bursty and bursty traffic

is the root of all evil…

• Shaping is highly beneficial if downstream device is policing

–Avoids the “instantaneous congestion”

–Space the traffic to conform to traffic contract

• Packet bursts are queued instead of being dropped, quickly training TCP sources to send at the desired rate

• Resulting packet stream is “smoothed” and net throughput for bursty traffic is higher

Why Traffic Shaping?

The slide lists some of the reasons for Traffic Shaping

Trang 20

econ_0481_09_010.ppt

Token Bucket

are added every Tc

at any time t during this period, upper bound being the access speed.

Over any integral multiple of Tc, the average bit rate of the interface will not exceed the mean bit rate The bit rate may, however, be arbitrarily fast

at any time t during this period, upper bound being the access speed.

In the token bucket metaphor, tokens are put into the bucket at a certain rate, Burst

Capacity (Bc) tokens every Time Interval Constant (Tc) seconds The bucket itself has

a specified capacity If the bucket fills to capacity (Bc + Excess Burst Capacity (Be)), newly arriving tokens are discarded Each token is permission fo r the source to send a certain number of bits into the network To send a packet, the regulator must remove

from the bucket a number of tokens equal in representation to the packet size

If not enough tokens are in the bucket to send a packet, the packet either waits until the bucket has enough tokens or the packet is discarded If the bucket is already full of

tokens, incoming tokens overflow and are not available to future packets Thus, at any time, the largest burst a source can send into the network is roughly proportional to the size of the bucket

Note that the token bucket mechanism used for traffic shaping has both a token bucket and a data buffer, or queue; if it did not have a data buffer, it would be a policer For

traffic shaping, packets that arrive that cannot be sent immediately are delayed in the

data buffer

For traffic shaping, a token bucket permits burstiness but bounds it It guarantees that

the burstiness is bounded so that the flow will never send faster than the token bucket's capacity plus the time interval divided by the established rate at which tokens are

placed in the bucket It also guarantees that the long-term transmission rate will not

exceed the established rate at which tokens are placed in the bucket

Bc is known as burst capacity Be is excess burst capacity Tc is the time interval

constant CIR is the Committed Information Rate All these terms are from

Frame-Relay

Trang 21

econ_0481_09_010.ppt

Token Bucket with Class-Based

Weighted Fair Queuing

• While (token bucket is not empty):

–De-queue traffic from Weighted Fair Queuing (WFQ)or traffic arrives (if WFQ empty)

–If token bucket is not empty:

• Token bucket = token bucket less message size

• Forward the traffic

–Else: fair queue the traffic

A token bucket is a formal definition of a rate of transfer It has three components:

a burst size, a mean rate, and a time interval (Tc) Although the mean rate is

generally represented as bits per second, any two values may be derived from the

third:

• Mean rate—Also called the committed information rate (CIR), it specifies how much data can be sent or forwarded per unit time on average

• Burst size—Also called the Committed Burst (Bc) size, it specifies in bits per

burst how much can be sent within a given unit of time to prevent scheduling

concerns

• Time interval—Also called the measurement interval, it specifies the time

quantum in seconds per burst

By definition, over any integral multiple of the interval, the bit rate of the interface will not exceed the mean rate The bit rate may, however, be arbitrarily fast within

the interval

Trang 22

econ_0481_09_010.ppt

Transmit Queue Output Line

Traffic Destined for Interface

Classification by:

Extended Access List Functionality

“Leaky Bucket”

Shaping

Configured Queuing (WFQ,

PQ, and so on)

Match

No Match Classify

(Generic) Traffic Shaping

Traffic shaping allows the control of traffic going out an interface in order to match its flow to the speed of the remote, target interface and to ensure that the traffic conforms to policies contracted for it Thus, traffic adhering to a particular profile can be shaped to meet

downstream requirements, thereby eliminating bottlenecks in topologies with

data-rate mismatches

The primary reasons traffic shaping should be used are to control access to available

bandwidth, to ensure that traffic conforms to the policies established for it, and to regulate the flow of traffic in order to avoid congestion that can occur when the sent traffic exceeds the access speed of its remote, target interface

Traffic shaping limits the rate of transmission of data Limit the data transfer to one of the following:

• A specific configured rate

• A derived rate based on the level of congestion

Generic Traffic Shaping (GTS) shapes traffic by reducing outbound traffic flow to avoid congestion by constraining traffic to a particular bit rate using the token bucket mechanism GTS applies on a per- interface basis and can use access lists to select the traffic to shape It works with a variety of Layer 2 technologies, including Frame Re lay, ATM, Switched

Multimegabit Data Service (SMDS), and Ethernet

On a Frame Relay subinterface, GTS can be set up to adapt dynamically to available

bandwidth by integrating Backward Explicit Congestion Notification (BECN) signals, or set

up simply to shape to a pre-specified rate GTS can also be configured on an ATM AIP model interface to respond to Resource Reservation Protocol (RSVP) signaled over statically

configured ATM permanent virtual circuits (PVCs)

Trang 23

• Simply forwards traffic

If not within threshold:

• Queues using WFQ-like queue on sub-interface

GTS is supported on most media and encapsulation types on the router GTS can

also be applied to a specific access list on an interface

Use DTS (covered in later slides) with the VIP cards

Trang 24

econ_0481_09_010.ppt

Traffic Destined for Interface

Queueing Match

No Match

Distributed Traffic Shaping

Enforces a maximum transmit rate Temporarily reduces transmit rate when signaled by Frame Relay (FR) Backward Explicit Congestion Notification (BECN) bits set in incoming frames

Shapes up to 200 FR Virtual Channels (VCs) at OC-3 rates with average size packets on a VIP2-50

Released in 12.0(4)XE, 12.0(7)S

Distributed Traffic Shaping (DTS) benefits:

• Offloads traffic shaping from the route switch processor (RSP) to the Versatile

Interface Processor (VIP)

• Supports up to 200 shape queues per VIP, supporting up to OC-3 rates when the

average packet size is 250 bytes or greater and when using a VIP2-50 or better with 8

MB of SRAM Line rates below T3 are supported with a

VIP2-40

The limitations are:

• Only IP traffic can be shaped

• dCEF must be enabled

• FastEtherChannel, Tunnel, VLAN and ISDN / Dialer interfaces are not supported

http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/120newft/120limit/120xe/120xe5/dts.htm

Trang 25

econ_0481_09_010.ppt

Frame-Relay Traffic Shaping

• Rate enforcement on a per-VC basis

–Peak rate for outbound traffic can be set to match CIR or another value

• Dynamic traffic throttling on a per-VC basis

–When BECN packets indicate congestion on the network, outbound traffic rate automatically stepped down

• Enhanced queuing support on a per-VC basis

–Custom queuing or priority queuing can be configured for individual VCs

• Can use different VCs for different types of traffic

FRTS provides these capabilities:

• Rate enforcement on a per-VC basis—the peak rate for outbound traffic The value can be set to match CIR or another value

• Dynamic traffic throttling on a per-VC basis—When BECN packets indicate congestion on the network, the outbound traffic rate is automatically stepped down; when congestion eases, the outbound traffic rate is increased This feature is enabled by default

• Enhanced queuing support on a per-VC basis—Either custom queuing or priority queuing can be configured for individual VCs

By defining separate VCs for different types of traffic and specifying queuing and an outbound traffic rate for each VC, bandwidth for each type of traffic is guarenteed By specifying

different traffic rates for different VCs over the same line, virtual time-division multiplexing is performed By throttling outbound traffic from high-speed lines in central offices to lower-speed lines in remote locations, congestion and data loss in the network is eased Enhanced queuing also prevents congestion-caused data loss

Trang 26

econ_0481_09_010.ppt

GTS

Shaper Interface Level or Group -Based

Shaping Queue WFQ

No Support for FRF.12 Understands BECN/FECN

FRTS

Shaper FR Only

Per DLCI

Shaping Queue PQ,CQ and WFQ(12.0(4)T)

Supports FRF.12 Understands FECN/BECN

DTS

Shaper on VIP Interface / subint Level

or Group-Based

Understands FECN/BECN

Shaping Queue WFQ

No Support for FRF.12

Difference Between FRTS, DTS, and GTS

Cisco has long provided support for forward explicit congestion notification (FECN) for

DECnet and OSI, and BECN for Systems Network Architecture (SNA) traffic using LLC2 encapsulation via RFC 1490 and DE bit support

FRTS builds upon this existing Frame Relay support with additional capabilities that improve the scalability and performance of a Frame Relay network, increasing the density of virtual circuits and improving response time

As is also true of GTS, FRTS can eliminate bottlenecks in Frame Relay networks that have high-speed connections at the central site and low-speed connections at branch sites Configure rate enforcement – a peak rate configured to limit outbound traffic – to limit the rate at which data is sent on the VC at the central site

Using FRTS, configure rate enforcement to either the CIR or some other defined value such as the excess information rate, on a per-VC basis The ability to allow the transmission speed used

by the router to be controlled by criteria other than line speed (that is, by the CIR or the excess information rate) provides a mechanism for sharing media by multiple VCs Pre-allocate

bandwidth to each VC, creating a virtual time-division multiplexing network

Also define PQ, CQ, and WFQ at the VC or subinterface level Using these queuing methods allows for finer granularity in the prioritization and queuing of traffic, providing more control over the traffic flow on an individual VC If CQ is combined with the per-VC queuing and rate enforcement capabilities, the Frame Relay VCs are enabled to carry multiple traffic types such

as IP, SNA, and Internetwork Packet Exchange (IPX) with bandwidth guaranteed for each traffic type

Trang 27

econ_0481_09_010.ppt

Summary Policy/Shaping

After completing this module section, you should be able to perform the following tasks:

• Describe the difference between policing and shaping and how each one relates to QoS

• Describe CAR, when to apply CAR, how to configure CAR

• Describe MQC policing and how to configure it

• Identify the three types of traffic shaping, their differences, and how to apply each

Trang 28

econ_0481_09_010.ppt

Section 2 Queuing

Trang 29

• Apply each queuing method for the right application

• Describe the difference between IP Rapid Transport Protocol (RTP) Priority and Low Latency Queuing (LLQ)

The purpose of the lesson is to quickly survey the queuing features in Cisco IOS

12.1, and to be able to describe the problems they solve

Trang 30

econ_0481_09_010.ppt

Application Traffic

App Tier 1 Tier 2 Tier 3 Tier 4

Market Data

256 kbps per router

128 kbps per pvc

96 kbps per router

120 kbps per pvc

96 kbps

on 1 pvc 48 kbps

SNA 56 kbps 48 kbps 36 kbps 24 kbps

Other (WWW, Email)

628 kbps 480 kbps

Stock brokerage customer needs to:

• Give priority to market data

• Reserve bandwidth by application

• Provide low latency for VoIP and SNA

How?

Descriptions of the classes of traffic and policy for this customer:

• Market Data—TCP Unicast, Top Priority, Bursty

• Hoot N Holler—VoIP Multicast, Sensitive to Drop, Jitter, and excessive drop

• Systems Network Architecture (SNA)—Delay sensitive Encapsulated in TCP

• Other—Delay insensitive, drop insensitive

Market Data is seen as the most important application The customer does not want anything to supercede it in importance or priority

The key is to know what traffic is going over each circuit Evaluate the traffic over

a period of time so that the location of QoS can be applied

Trang 31

Marked (colored) at ingress at network edge

IP Precedence / Differentiated Services Code Point (DSCP) set Policed and shaped at the network edge

Potentially discarded by Weighted Random Early Detection (WRED) (congestion mgmt)

Assigned to the appropriate outgoing queue Scheduled for transmission by CBWFQ and sent out interface

Classify and Mark

Trang 32

Advanced queuing methods Scheduling

Sequence of events

Basic queuing methods are examined first

One of the most basic (not shown) is FIFO, or First-In First-Out queuing This is

the technique used in simple routers and (generally) in lines of people No special

treatment for anyone!

Generally use a queuing technique when there is congestion, or not enough

capacity for all the traffic Most queuing descriptions focus on which traffic gets

preferential treatment However, if there is congestion, some traffic will have to be dropped So all queuing schemes, implicitly, are also making a choice as to which

Trang 33

econ_0481_09_010.ppt

Priority Queuing

• Rigid traffic prioritization scheme with four queues—

high, medium, normal, low

• Traffic assigned to queues by previously defined priorities and policies

• Unclassified packets are placed in the normal queue

Interface Buffer Resources

High Medium Normal Low Classify

Absolute Priority Scheduling

The disadvantage of priority queuing is that the higher queue is given absolute

precedence over lower queues For example, packets in the low queue are only sent when the high, medium, and normal queues are completely empty If a queue is

always full, the lower-priority queues are never serviced They fill up and packets

are lost Thus, one particular kind of network traffic can dominate a priority

queuing interface

An effective use of priority queuing would be to place time-critical but

low-bandwidth traffic in the high queue This ensures that this traffic is sent

immediately, but because of the low-bandwidth requirement, lower queues are

unlikely to be starved

In order for packets to be classified on a priority queuing interface, create policies

on that interface These policies need to filter traffic into one of the four priority

queues Any traffic that is not filtered into a queue is placed in the normal queue

For additional informaton refer to:

http://cisco.com/univercd/cc/td/doc/product/software/ios121/121cgcr/qos_c/qcprt2/qcdpq.htm

Trang 34

Up to 16

3 1

Weighted Round Robin Scheduling (byte count)

Allocate Proportion of Link Bandwidth

Classify

2 5 4

Transmit Queue

Output Line

The disadvantage of custom queuing is that, as in priority queuing, policy statements on the interface must be created to classify the traffic to the queues An effective use of custom queuing would be to guarantee bandwidth to a few critical applications to ensure reliable application performance

In order for packets to be classified on a custom queuing interface, create custom

queuing policies on that interface These policies need to specify a ratio, or percentage,

of the bandwidth on the interface that should be allocated to the queue for the filtered traffic A queue percentage can be as small as 5%, or as large as 95%, in increments of 5% The total bandwidth allocation for all policy statements defined on a custom

queuing interface cannot exceed 95% Any bandwidth not allocated by a specific policy statement is available to the traffic that does not satisfy the filters in the policy

statements

The queues that are defined constitute a minimum bandwidth allocation for the specified flow If more bandwidth is available on the interface due to a light load, a queue can use the extra bandwidth This is handled dynamically by the device

All packets that have not been classified for custom queuing are placed in the default queue

Figuring out how to allocate bandwidth can require either some fairly simple theoretical bandwidth calculations or a pragmatic (try-then-tweak) approach, in order to come up with the right byte-threshold limits for the custom queues

For additional information refer to:

http://cisco.com/univercd/cc/td/doc/product/software/ios121/121cgcr/qos_c/qcprt2/qcdcq.htm

Trang 35

econ_0481_09_010.ppt

Weighted Fair Queuing

• Traffic automatically sorted by stream

over high-bandwidth traffic

interfaces

WFQ provides traffic priority management that automatically sorts among individual traffic streams without requiring that access lists be defined first WFQ can also manage duplex data streams such as those between pairs of applications, and simplex data streams such as voice or video In effect, there are two categories of WFQ sessions: high bandwidth and low bandwidth Low-bandwidth traffic has effective priority over high-bandwidth traffic, and high-bandwidth traffic shares the transmission service proportionally according to assigned weights

When WFQ is enabled for an interface, new messages for high-bandwidth traffic streams are discarded after the configured or default congestive messages threshold has been met However, low-bandwidth conversations, which include control message conversations, continue to queue data As a result, the fair queue may occasionally contain more messages than its configured threshold number specifies

With standard WFQ, packets are classified by flow Packets with the same source IP address, destination IP address, source TCP or UDP port, or destination TCP or UDP port belong to the same flow WFQ allocates an equal share of the bandwidth to each flow Flow-based WFQ is also called fair queuing because all flows are equally weighted

WFQ is the default queuing mode on interfaces that run at or below E1 speeds (2.048 Mbps or less) It is enabled by default for physical interfaces that do not use Link Access Procedure,

Trang 36

Versatile Interface Processors (VIPs)

• Available on Cisco 7000 routers with RSP7000 or 7500 routers with VIP2-40 or faster

For dWFQ, packets are classified by flow Packets with the same source IP

address, destination IP address, source TCP or UDP port, destination TCP or UDP port, and protocol belong to the same flow

dWFQ can be configured on interfaces but not sub- interfaces It is not supported on Fast EtherChannel, tunnel, or other logical or virtual interfaces such as MLP

To monitor WFQ or dWFQ, use the following commands:

• show interfaces [interface] fair-queue

• show queueing fair

Example:

Router# show interfaces hssi 0/0/0 fair-queue

Hssi0/0/0 queue size 0

packets output 35, drops 0

WFQ: global queue limit 401, local queue limit 200

For additional details and configuration commands for both WFQ and dWFQ see

the following URL:

http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/12cgcr/qos_c/qcpart2/qcwfq.htm

Trang 37

econ_0481_09_010.ppt

Weighted Random Early Detection

Upon congestion, packets from lower precedence are selectively discarded first Minimize the congestion impact on higher precedence services

Standard Premium

Weighted Random Early Detection (WRED) is a congestion-avoidance mechanism

to prevent congestion, not manage it WRED monitors traffic load on an interface

It selectively discards lower-priority traffic when the interface starts to get

congested

For robust transport protocols (such as TCP), WRED has the effect of throttling

back lower priority traffic at the source TCP has built- in rate shaping mechanisms,

so when congestion occurs the applications will throttle back

WRED, CQ, PQ, and WFQ are mutually exclusive on an interface The router

software produces an error message if WRED and any one of these queuing

strategies is configured simultaneously

WRED is well- suited for avoiding congestion on high-speed backbone links

When testing QoS with a SmartBits analyzer, bear in mind that the SmartBits does not respond to packet drops, hence the testing will not accurately reflect the

benefits of WRED

Trang 38

econ_0481_09_010.ppt

A Possible Implementation of Assured Forwarding with WRED

Different drop preference for different precedence

th1 th2 th3Transmit

Probability

of drop increases with increasing average queue size

AF Class

The slide shows how WRED works Between initial threshold and queue size, the

probability of dropping a packet increases When the queue is full, the next packet

must always be dropped

WRED helps break up a TCP/IP phenomenon known as porpoising This is where

congestion avoidance in TCP causes a number of flows to all reduce their send

windows at the same time It is well known that TCP flows do tend to

synchronization of send windows and group slowdowns (similar to a school of

porpoises surfacing in the ocean at one time) It is better to randomly sacrifice a

packet from one flow so that many flows do not unnecessarily slo w down at the

same time

The WRED mechanism gives a possible implementation of DiffServ Assured

Forwarding (AF)

Trang 39

econ_0481_09_010.ppt

Flow RED

example, User Datagram Protocol (UDP))

all the buffer resources

interface Serial1 random-detect random-detect flow random-detect flow average -depth- factor 8 random-detect flow count 16

FRED tends to penalize non-adaptive flows A flow is considered non-adaptive, that is, it takes up too much of the resources, when the average flow depth times the specified

multiplier (scaling factor) is less than the flow's depth, that is: average-flow-depth *

(scaling factor) < flow-depth For test comparisons, FRED penalizes the non-adaptive

SmartBits tester more heavily It is also “too strict” for use with UDP voice traffic

To enable flow-based WRED, also known as FRED, use the random-detect flow interface

configuration command: random-detect flow

Use this command to enable flow-based WRED before using the random-detect flow

average-depth-factor and random-detect flow count commands to further configure the

parameters of flow-based WRED For additional information, refer to:

http://cisco.com/univercd/cc/td/doc/product/software/ios121/121cgcr/qos_r/qrdcmd4.htm#xtocid79759

To set the multiplier to be used in determining the average depth factor for a flow when

FRED is enabled, use the random-detect flow average-depth-factor interface

configuration command: random-detect flow average-depth-factorscaling-factor

Use this command to specify the scaling factor that flow-based WRED should use in

scaling the number of buffers available per flow and in determining the number of packets allowed in the output queue for each active flow This scaling factor is common to all

Trang 40

Tiêu đề	Policy / Shaping
Trường học	Cisco Systems, Inc.
Chuyên ngành	Wide Area Networks
Thể loại	Tài liệu
Năm xuất bản	2000
Thành phố	San Jose

Định dạng
Số trang	82
Dung lượng	2,17 MB