Traffic Shaping and Policing

IP QoS Traffic Shaping and Policing- 5Traffic Shaping and Policing Traffic Shaping and Policing • Traffic Shaping and Policing mechanisms are used to rate-limit traffic classes • They ha

Trang 1

It includes the following topics:

Objectives

Upon completion of this module, you will be able to perform the following tasks:

(Class-based Policing and Class-(Class-based Shaping)

Trang 2

Traffic Shaping and Policing

Overview

The lesson introduces mechanisms for traffic policing and traffic shaping

Committed Access Rate (CAR), Generic Traffic Shaping (GTS) and Frame Relay Traffic Shaping (FRTS) are introduced in this section

Objectives

Upon completion of this lesson, you will be able to perform the following tasks:

mechanisms

Trang 3

• Traffic Shaping and Policing mechanisms are used to rate-limit traffic classes

• They have to be able to classify packets and meter their rate of arrival

• Traffic Shaping delays excess packets to stay within the rate limit

• Traffic Policing typically drops excess traffic to stay within the limit; alternatively it can remark excess traffic

Meter

Traffic stream

Both shaping and policing mechanisms are used in a network to control the rate at which traffic is admitted into the network Both mechanisms use classification, so they can differentiate traffic They also use metering to measure the rate of traffic and compare it to the configured shaping or policing polic y

The difference between shaping and policing can be described in terms of their rate-limiting implementation:

within the desired rate limit With shaping, traffic bursts are smoothed out producing a steadier flow of data Reducing traffic bursts helps reduce congestion in the core of the network

limits Policing does not introduce any delay to traffic that conforms to traffic policies It can however, cause more TCP retransmissions, because traffic in excess of specified limits is dropped

Trang 4

Why Use Rate Limiting

network with asymmetric link bandwidths

• To limit access to resources when speed access is used but not desired

high-• To limit certain applications or classes

Rate limiting is typically used to satisfy one of the following requirements:

asymmetric bandwidths are used along the traffic path This prevents the layer-2 network from dropping large amounts of traffic by differentiately dropping excess traffic at ingress to the ATM or Frame Relay networks based

on Layer-3 information (for example: IP precedence, DSCP, access list, protocol type, etc.)

is used in transport, but sub-rate access is desired

traffic follow a specified traffic -rate policy

bandwidth characteristics of a TDM system (that is, fixed maximum available bandwidth) Inbound and outbound policing can, for example, be used on one router to split a single point-to-point link into two or more virtual point-to-point links by assigning a portion of the bandwidth to each class, thus preventing any class from monopolizing the link in either direction

Trang 5

Typical Traffic Shaping or Policing Applications

Low-speed link High-speed

link

Output interface is not congested queuing and WRED

do not work

Congestion in WAN network results in non-intelligent layer-

2 drops

Server Farm

Implementing a virtual TDM or Leased line over a single physical link

high-2 network

The second picture shows a hosting farm, which is accessible from the Internet via

a shared link Depending on the service contract, the hosting provider may offer different bandwidth guarantees to customers, and may want to limit the resources

a particular server uses Rate limiting can be used to divide the shared resource (upstream link) between many servers

The third example shows the option of implementing virtual leased lines over a Layer-3 infrastructure, where rate-limited reserved bandwidth is available over a shared link

Trang 6

Shaping vs Policing

• Benefits of Shaping

– Shaping does not drop packets

– Shaping supports interaction with Frame Relay congestion indication

• Benefits of Policing

– Policing supports marking

– Less buffer usage (shaping requires an additional queuing system)

A shaper typically delays excess traffic using a buffer, or mechanism, to hold packets and shape the flow when the data rate of the source is higher than expected Traffic shaping smoothes traffic by storing traffic above the configured rate in a queue Therefore, shaping increases buffer utilization on a router, but causes non-deterministic packet delays Shaping can also interact with a Frame Relay network, adapting to indications of Layer-2 congestion in the WAN

A policer typically:

packets in needed)

Both policing and shaping ensure that traffic does not exceed a bandwidth limit, but they have different impacts on the traffic:

connection-oriented protocols

Trang 7

How do Routers Measure Traffic

configured rate limit

The metering is usually performed with an abstract model called a token bucket, which is used when processing each packet The token bucket can calculate whether the current packet conforms or exceeds the configured rate limit on an interface

Trang 8

700

Token Bucket

500 bytes Conform Action 500 bytes

The token bucket is a mathematical model used in a device that regulates the data flow The mode has two basic components:

bits into the network

Tokens are put into the bucket at a certain rate by the operating system Each incoming packet, if forwarded, takes tokens from the bucket, representing the packet’s size

If the bucket fills to capacity, newly arriving tokens are discarded Discarded tokens are not available to future packets

If there are not enough tokens in the bucket to send the packet, the regulator may:

The figure shows a token bucket, with the current capacity of 700 bytes When a 500-byte packet arrives at the interface, its size is compared to the bucket capacity (in bytes) The packet conforms to the rate limit (500 bytes < 700 bytes), and the packet is forwarded 500 tokens are taken out of the token bucket leaving 200

Trang 9

in the case of shaping)

Trang 10

Token Bucket

• Bc is normal burst size (specifies sustained rate)

• Be is excess burst size (specifies length of burst)

Bc+ Be

B c of tokens is added every T c [ms]

T c = B c / CIR

Time

Link Utilization

T c 2*T c 3*T c 4*T c 5*T c

B c B c B c B c B c B c

Link BW

Average BW (CIR)

B e

CIR is the Committed Information Rate (also called the committed rate, or the

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

tokens are discarded Each token grants permission for a source to send a certain number of bits into the network To send a packet, the regulator must remove, from the bucket, the number of tokens equal in representation to the packet size For example, if 8000 bytes worth of tokens are placed in the bucket every 125 milliseconds, the router can steadily transmit 8000 bytes every 125 milliseconds, if traffic constantly arrives at the router

If there is no traffic at all, 8000 bytes per 125 milliseconds get accumulated in the

collects 64000 bytes worth of tokens, which can be transmitted immediately in the

which can be transmitted in a single burst, at the line rate

Trang 11

Although token bucket permits burstiness, traffic bursts are bound This guarantee

is made so that traffic flow will never send faster than the token bucket's capacity

In the long-term, this means that the transmission rate will not exceed the established rate at which tokens are placed in the bucket (the committed rate)

Trang 12

Mechanisms

– Generic Traffic Shaping (GTS)

– Frame Relay Traffic Shaping (FRTS)

Two methods are policing mechanisms:

All these methods are discussed next in specific sections

Trang 13

Summary

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

mechanisms

Lesson Review

Answer the following questions:

1 How do shaping and policing mechanisms keep track of the traffic rate?

2 Which shaping mechanisms are available with the Cisco IOS software?

3 Which policing mechanisms are available with the Cisco IOS software?

4 What are the main differences between shaping and policing?

Trang 14

Generic Traffic Shaping

Overview

This lesson describes the Generic Traffic Shaping (GTS) mechanism

Objectives

Upon completion of this lesson, you will be able to perform the following tasks:

Trang 15

Generic Traffic Shaping

• Can shape multiple classes ( classification )

• Can measure traffic rate of individual classes ( metering )

( shaping )

Traffic stream

Dropper Meter

Generic Traffic Shaping (GTS) shapes traffic by reducing the outbound traffic flow

to avoid congestion This is achieved by constraining traffic to a particular bit rate using the token bucket mechanism GTS is applied 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 Relay, ATM, Switched Multi-megabit Data Service (SMDS) and Ethernet

As shown in the block diagram, GTS performs three basic functions:

policies applied to them

and exceeding traffic

configured rate limit

Trang 16

Yes

Shaping WFQ

Arriving packets are first classified into one of the shaping classes Traffic not classified into any class is not shaped Classification can be performed using access lists

Once a packet is classified into a shaping class, its size is compared to the amount

of available token in the token bucket of that class The packet is forwarded to the main interface queue if there are enough tokens A number of tokens taken out of the token bucket is equal to the size of the packet (in bytes)

If, on the other hand, there are not enough tokens to forward the packet, the packet is buffered in the WFQ system assigned to this shaping class The router will then periodically replenish the token bucket and check if there are enough tokens to forward one or more packets out of the shaping queue Packets are scheduled out of the shaping queue according to the WFQ scheduling algorithm

Trang 17

GTS Overview

• GTS is multiprotocol

any queuing mechanisms:

– FIFO Queuing

– Priority Queuing (PQ)

– Custom Queuing (CQ)

– Weighted Fair Queuing (WFQ)

The GTS implementation in Cisco IOS supports multiple protocols and works on a varie ty of interface types WFQ is used as the shaping delay queue, providing fair scheduling within a traffic class Other queuing strategies (FIFO, PQ, CQ and WFQ) may be employed after GTS to provide traffic scheduling on the shaped traffic Also, GTS only works at the output of an interface

GTS can be used to shape all outbound traffic on an interface or it can separately shape multiple classes Classification is performed using any type of access list including all non-ip access lists

Trang 18

GTS Implementation

• The software queue may have no function if the sum of all shaping rates is less than link bandwidth

Shaping Queue (WFQ)

Software Queue (FIFO, PQ,

CQ, WFQ, )

Hardware Queue (FIFO)

Dispatches packets at configured rate

Dispatches packets at line rate

Bypass the software queue

if it is empty and there is room in the hardware queue

Packet flow through GTS is implemented using three queues The first, the shaping queue, is WFQ-based and shapes traffic according to the specified rate using a token bucket model This queue dispatches packets to the software queue, which may be configured with other queuing mechanisms (PQ, CQ, WFQ or FIFO) If the software queue is empty, traffic is forwarded directly to the output hardware queue

GTS supports distributed implementation on VIP adapters This offloads traffic shaping from the route switch processor (RSP) to the Versatile Interface Processor (VIP), and constructs all of the queues in VIP packet memory Only IP traffic can be shaped with dWFQ Another requirement is that dCEF switching must be enabled

Trang 19

traffic-shape rate bit-rate [burst-size

[excess-burst-size]]

Router( config-if)#

To enable traffic shaping for outbound traffic on an interface, use the

traffic-shape rate interface configuration command Of the parameters to be specified,

bit-rate is the only mandatory one The burst-size and excess-burst-size are optional

Generic traffic shaping can be used in all switching paths Older Cisco IOS versions may use slower switching paths when GTS is in effect

Trang 20

Default value: 1/8 of bit rate

Router(config -if)#

Bit rate (in bits per second) is configured as the average traffic rate to which the

traffic should be shaped on the output of the interface

Burst size (in bits) can be configured to allow for varying levels of allowed

burstiness That is, traffic, which bursts over the average traffic rate, also conforms if it falls within the burst rate in an interval By default, this is set to one

Trang 21

parameter, defines the excess burst of traffic, which can still be sent through the first noticed burst By default, there is no excess burst allowed

of the token bucket By default, it is directly computed from the bit rate and the

shaping, those parameters are bounded to values between 25 and 125 ms

Trang 22

Configuring GTS

• Shapes outbound traffic matched by the specified access list

• Several traffic-shape group commands can be configured on

the same interface

• The “ traffic-shape rate “ and “ traffic-shape group “ commands

cannot be mixed on the same interface

• Separate token bucket and shaping queue is maintained for

each traffic-shape group command

• Traffic not matching any access list is not shaped

traffic-shape group access-list bit-rate [burst

use the traffic-shape group interface configuration command The traffic-shape

group command allows specification of one or more previously defined access

lists to shape traffic on the interface One traffic-shape group command must be

specified for each access list on the interface

Cisco IOS uses separate token buckets and shaping queues for each class, as differentiated by the access list specification Traffic not matching any access list bypasses traffic shaping and is immediately sent to the software or hardware interface queue

Use the traffic-shape rate command if no classification is needed and shaping should be applied to all traffic Remember that the traffic-shape group command using an IP access list permitting all IP traffic is not equivalent to the traffic-shape

rate command if non-IP traffic is present in the network

Trang 23

GTS Example #1

• ISP wants to sell a service in which a customer may use all of a E1 line for 30 seconds in a burst, but on a long term average is limited to 256 kbps

msec

In the first GTS example, an ISP wants to control the amount of traffic injected into the Frame Relay WAN by the customer The SP service uses an E1 line as the access line, limits the customer to 256 Kbps on the average, but also permits bursts of up to thirty seconds at the E1 line rate

The parameters are calculated based on the service requirements CIR (the average bit rate) is set at the specified average rate, the burst size is set to one eighth of the CIR (32000 bits), and the excess burst size reflects the allowed thirty-second burst at full E1 line rate

The excess burst size was calculated using the following formula:

1 Each second of transmission at line-speed requires 2 Mbits

2 Thirty second burst therefore requires 30 x 2 Mbits

3 The excess burst size is 30 x 2048000 = 61440000

It takes thirty seconds to empty the token bucket How long does it take to fill it up again?

The token bucket is emptied at 2Mbps but it is replenished at 256kbps It takes eight times as long to fill it as it does to empty it Every thirty second burst would, therefore, require a four-minute silence on the line to accumulate tokens

Trang 24

Core Customer

GTS Example #1

interface ethernet 0/0 traffic -shape rate 256000 32000 61440000

! interface serial 1/0 traffic -shape rate 256000 32000 61440000

interface ethernet 0/0 traffic-shape rate 256000 32000 61440000

! interface serial1/0 traffic-shape rate 256000 32000 61440000

the configuration would be done on both the inbound and outbound interfaces

WAN

The figure shows the router configuration required to implement this service All the output traffic is shaped, and the shaping needs to be configured on all customer edge sites, which will perform admission control using GTS

Trang 25

Core Customer

GTS Example #2

traffic will never use more than 64 kbps

WAN

interface ethernet 0/0 traffic-shape group 101 64000 interface serial 1/0

traffic-shape group 101 64000

! access -list 101 permit tcp any any eq www

interface ethernet 0/0 traffic -shape group 101 64000 interface serial 1/0

traffic -shape group 101 64000

! access-list 101 permit tcp any any eq www

In the second example, a customer wants to limit web usage, so that web traffic never uses more than 64 Kbps on the access link The router configuration is shown in the figure, using default parameters for traffic bursts An access list defines web traffic as the only shaped traffic All other traffic bypasses GTS and can use the full access line bandwidth

Trang 26

Monitoring GTS

Router#show traffic-shape access Target Byte Sustain Excess Interval Increment Adapt I/F list Rate Limit bits/int bits/int (ms) (bytes) Active Se3/3 100000 2000 8000 8000 80 1000 -

Router# show traffic-shape

access Target Byte Sustain Excess Interval Increment Adapt I/F list Rate Limit bits/int bits/int (ms) (bytes) Active

show traffic-shape

Router(config )#

The figure shows the results of the show traffic-shape command issued on a

To display the current traffic-shaping configuration, use the show traffic-shape command To display the current traffic -shaping statistics, use the show traffic-

shape statistics command Output of both the commands is detailed in the

ensuing figures

Information displayed includes:

by the CIR)

Trang 27

Monitoring GTS

Router# show traffic -shape statistic Access Queue Packets Bytes Packets Bytes Shaping I/F List Depth Delayed Delayed Active Se3/3 77 16091 3733112 414 96048 yes

Router# show traffic-shape statistic

Access Queue Packets Bytes Packets Bytes Shaping I/F List Depth Delayed Delayed Active Se3/3 77 16091 3733112 414 96048 yes

Depth of the associated WFQ queue for delayed packets

Number of packets/bytes sent

on the interface

Subset of the previous number of packets/bytes delayed via the WFQ queue

show traffic-shape statistic

Router(config )#

The show traffic-shape statistics command displays the statistics of traffic

shaping for all the configured interfaces Displayed in the output is:

is used (traffic-shape rate command is used on interface serial3/3 in the

example)

command since the last clearing of interface counters (16091 packets in the example)

command since the last clearing of interface counters (3733112 bytes in the example)

command since the last clearing of interface counters (414 packets in the example)

command since the last clearing of interface counters (96048 bytes in the example)

The expected result of traffic shaping is a high ratio between transmitted packets and delayed packets

Trang 28

If the number of delayed packets is very high (compared to the total number of packets) then there are probably non-responsive aggressive flows being shaped and the queue depth could show high buffer utilization

If the number of delayed packets is zero then it is very likely that the access list does not match any traffic

Trang 29

Monitoring GTS

router#show traffic-shape queue Traffic queued in shaping queue on Serial0 (depth/weight) 1/4096

Conversation 254, linktype : ip, length: 232 source: 1.1.1.1, destination: 1.1.2.47, id: 0x0001, ttl: 208, TOS: 0 prot: 17, source port 11111, destination port 22222

router# show traffic -shape queue

Traffic queued in shaping queue on Serial0 (depth/weight) 1/4096

Conversation 254, linktype: ip, length: 232 source: 1.1.1.1, destination: 1.1.2.47, id: 0x0001, ttl: 208, TOS: 0 prot: 17, source port 11111, destination port 22222

• Displays the shaping queue contents

show traffic-shape queue

Router( config)#

The show traffic-shape queue command displays the contents of the shaping

queue associated with an interface

This command can be used to determine the types of flows that are congesting the shaping queue The command displays the parameters that are used for

The example shows that there is a non-responsive UDP flow (protocol 17) congesting the shaping queue

Trang 30

GTS on Frame Relay Interfaces

(sub)interface

implemented on Frame Relay interfaces:

– Adaptation to Frame Relay congestion notification

– BECT-to-FECN reflection

– FECN creation on congestion

GTS applies on a per-interface basis, can use access lists to select the traffic to shape, and works with a variety of Layer-2 technologies, including:

Trang 31

Frame Relay Refresher

• Frame Relay Explicit Congestion Notification – FECN (Forward Explicit Congestion Notification)

– BECN (Backward Explicit Congestion Notification)

– CLLM (Consolidated Link Layer Management)

• Implicit Congestion Notification

– Network discards detected by end user at higher layers

– DE (Discard Eligibility) bit

Frame Relay performs congestion notification to its Layer-2 endpoints by including congestion signaling inside the Layer-2 frame headers

in-band congestion signaling

Relay network to notify a device (FR DTE, which may be a router) that it should initiate congestion avoidance procedures

Relay network to notify a device (DTE) that it should initiate proper congestion avoidance procedures

expands on the FECN/BECN mechanism to improve congestion management

preference to other frames, if congestion occurs, to maintain the committed quality of service within the network Frames with the DE bit set are

Congestion notification may be explicit (honored by Layer-2 devices) or implicit (detected and honored by higher-layer protocols, not by the Layer-2 network) FECN/BECN and CLLM are explicit methods, while BE-setting is an implicit notification method

Trang 32

Frame 1 Frame 1 Frame 1 FECN FECN

Frame 2 Frame 2 BECN

Frame 2 BECN

Congestion this Side

No Congestion this Side

Switch monitors all transmit queues for congestion

S e n d e r

R e c e i v e r

Frame Relay Switch

Frame Relay FECN/BECN Congestion Control

Same Virtual Circuit (VC)

• FR Switch detects congestion on output queue and informs:

–The receiver by setting the FECN bit on forwarded frames

–The source by setting the BECN bit on frames going in the opposite direction

A Frame Relay switch can explicitly report congestion in two directions: Forward and Backward When a frame queue inside a switch is congested, the switch will generate congestion signals based on the FECN and BECN bits If congestion occurs in a queue towards the main receiver of traffic, FECN signals are sent to the receiving Layer-2 endpoint and BECN signals are sent to the sending Layer-2 endpoint FECN and BECN bits are not sent as separate frames, but are

piggybacked inside data frames

Trang 33

GTS Frame Relay Congestion

– The GTS bit rate is reduced when BECN packets are received to reduce the data flow through congested Frame Relay network

– Adaptation is done on per (sub)interface basis

– GTS bit rate is gradually increased when the congestion is no longer present (no BECN packets are received any more)

BECN is the flag that the sending DTE (router as a Frame Relay endpoint) is able

to integrate to determine the congestion status of the Layer-2 WAN

Trang 34

GTS Frame Relay Congestion Adaptability Mechanisms

• FECN to BECN propagation

– A test packet with BECN bit set is sent to the sender if a packet with FECN bit set is received

The first adaptation mechanism is bit-rate adaptation GTS is able to respond to Layer-2 congestion by reducing its shaping rate to three-quarters of the current rate, until the Layer-2 network recovers from congestion When BECN flags are

no longer received, the rate is slowly ramped up again to the original shaping rate This is also a lower limit of rate reduction, which bounds the reduction process so that at least some throughput is maintained The BECN-integrating functionality is performed on a per sub-interface (DLCI) basis

However, if the congestion was caused by simplex traffic (such as a multicast video stream) or by an aggressive TCP connection, it is expected that the reverse traffic (frames flowing from the receiver to the sender, marked with the BECN bit) might come by less frequently than required to feed the integration So the receiving DTE (the receiving router) can help matters when it receives a message with FECN set by first checking to see if it has any data, and if it does not,

originating a message with BECN set This message might be a Q.922 TEST RESPONSE message, which would by virtue of its message type be understood to

be a message to discard and not reply to This feature is called FECN-to-BECN propagation

Trang 35

An Example of BECN Integration

BECN Integration

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

1 2 3 4 5 6 7 8 9 10 1 1 12 1 3 14 15 1 6 17 1 8 19 20 21 22 2 3 24 25

time represented in units of Tc

Inc

becn becn

traffic-shape rate 64000 8000 8000 traffic-shape adaptive 32000 BECN received at Tc#1 and Tc#3 Hypothesis: no idle traffic

The figure shows the shaped rate of a token bucket-based GTS responding to BECN packets it received As mentioned, the rate is reduced to three-quarters of

shaped rate is brought up slowly, up one-sixteenth of the current rate

Trang 36

Congestion

FECN to BECN Propagation

S e n d e r

R e c e i v e r

If there is no reverse traffic, the switch is not able to set BECN in frames going back

to sender

BECN in Q.922Test

FECN

Frame Relay Switch

The other adaptation method, FECN-to-BECN propagation, configures a Frame Relay sub-interface to reflect received FECN bits as BECN in Q.922 TEST RESPONSE messages This enables the sender to notice congestion in the Layer-

2 network, even if there is no data traffic flowing from the receiver back to the sender

Trang 37

Configuring Bit-rate Adaptation

• Configures Traffic Shaping Frame Relay bit-rate adaptation

response to continuous BECN signals Default: 1/2 the specified traffic shaping rate

• Traffic shaping has to be enabled

traffic-shape adaptive [bit-rate]

Router(config-if)#

Frame Relay bit rate adaptation is configured using the traffic-shape adaptive

command, which specifies the lower limit to which the shaped rate should be reduced in presence of incoming BECN signals By default, this is half the configured sustained (committed) rate in GTS The bit rate is configured in bits per second

Trang 38

• Configures the router to send Frame Relay TEST message with BECN bit set in response to receiving

a frame with FECN bit set

• Can be used without adaptive traffic shaping

Configuring FECN to BECN

Router( config-if)#

The traffic-shape fecn-adapt command enables the FECN-to-BECN

propagation It can be used without adaptive GTS, as configured with the previous command

This feature should be used for testing purposes only If the feature is combined with the adaptation feature it is very likely that the first delayed packet will cause the shaping to slow down to the minimum shaping rate For example:

1 Router A (sender) sends a frame with a FECN bit because it had to delay

a packet

2 Router B (receiver) replies with the TEST frame with the BECN bit set

3 Router A (sender) reduces the shaping rate due to the received BECN causing even more delay and more packets with the FECN bit set

Trang 39

GTS Frame Relay Adaptation Design

Conservative scenario

• Set shaping rate to CIR

• Set minimum rate to MIR (or 1/2 CIR)

Optimistic scenario

• Set shaping rate to EIR

• Set minimum rate to CIR

Realistic scenario

• Set shaping rate to EIR

• Set minimum rate to MIR (or 1/2 CIR)

To illustrate different possibilities of adaptation, consider the following three scenarios for using GTS over a Frame Relay circuit

dropping, the shaping rate is set to the contracted Frame Relay CIR (Committed Information Rate) and the minimum rate of adaptation is set either

to MIR (Minimum Information Rate) or half the CIR value MIR depends on the provider’s over provisioning of the network and can be as low as one-tenth

of the CIR This configuration minimizes dropping, but does not allow excess bandwidth to be fully utilized

(Excess Information Rate) and the minimum rate to the CIR This configuration would probably cause too much dropping in a loaded Frame Relay network

setting the shaping rate to the EIR and the minimum adaptation rate to the MIR (or half the CIR) This would allow full advantage to be made of the Frame Relay network, if possible, and to adapt to a realistic level if congestion

is indicated

Trang 40

Core Customer

interface serial 0/0 traffic -shape rate 64000 8000 8000 traffic -shape adaptive 48000

Tiêu đề	Traffic Shaping and Policing
Chuyên ngành	Computer Networks
Thể loại	Giáo trình
Năm xuất bản	2001

Định dạng
Số trang	104
Dung lượng	4,69 MB