Upload
nguyenkiet
View
291
Download
0
Embed Size (px)
Citation preview
© 2013 Cisco and/or its affiliates. All rights reserved. Presentation_ID Cisco Public
Joining Audio Broadcast
2
1. Audio Broadcast window should automatically pop up; Audio will be streamed through your computer speakers
2. If Audio Broadcast window does not appear, go to Communicate menu and select Audio Broadcast
3. You will hear hold music until the event begins
4. If you are unable to hear via your PC speakers, click the Phone button to request dial-in instructions
© 2013 Cisco and/or its affiliates. All rights reserved. Presentation_ID Cisco Public
Submit Questions
3
1. Click on the Q&A arrow to expand the Q&A panel
2. Type your question Technical Assistance – send to Host Content Questions – send to All Panelists
3. Click the Send button
© 2013 Cisco and/or its affiliates. All rights reserved. Presentation_ID Cisco Public
Thank You for Joining Us Today
4
Download a copy of todays slides using the link in the chat.
Today’s webcast will be available on-demand within 48hrs.
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Your Presenters
5
Tim Szigeti
A Case Study in Strategic QoS Design
Tim Szigeti, CCIE# 9794 Senior Technical Leader—Systems Design Unit
Cisco Systems
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
A Case Study in Strategic QoS Design
§ This session will review QoS tools, best practices and design strategies
§ The goal of this session is to present the thought-process to formulate an end-to-end QoS strategy that meets current and future business needs
§ Practical application of these concepts and recommendations will be made through a case-study design example
Session Abstract
7
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
A Case Study in Strategic QoS Design
§ QoS Tools Review
§ QoS Design Best Practices
§ QoS Design Strategies
§ QoS Design Case Study
Agenda
8
QoS Tools Review
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Tools Review: Classification & Marking Tools
§ Classification: – An action that organizes packets into different traffic types, to which different policies
can then be applied – Classification of packets can happen without marking
§ Marking: – Writes a value into the packet header – Establishes a trust boundary at the network edge – Can be used in other locations in the network and is not always used solely for
purposes of classification
Classification vs. Marking
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Tools Review: Classification & Marking Tools
§ Classification can be done on: – Layer 1 criteria—such as ingress physical interface – Layer 2 criteria—such as IEEE 802.1Q/p CoS – Layer 3 criteria—such as IP DSCP – Layer 4 criteria—such as TCP/UDP port(s) – Layer 7 criteria—such as NBAR application signatures
§ Marking can be done on: – Layer 2 fields—such as IEEE 802.1Q/p CoS – Layer “2.5” fields—such as MPLS EXP – Layer 3 fields—such as IP DSCP – Internal fields—such as QoS Group
Classification and Marking Options
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
§ 802.1p User Priority field also called Class of Service (CoS)
§ Different types of traffic are assigned different CoS values
§ CoS 6 and 7 are reserved for network use
802.1Q 4 Bytes
Three Bits Used for CoS (802.1p Class of Service)
Data FCS PT SA DA SFD Pream Type
PRI VLAN ID CFI
Ethernet Frame
CoS Acronym Traffic characteristics
0 BE Best Effort 1 BK Background 2 EE Excellent Effort 3 CA Critical Applications 4 VI Video, < 100 ms latency 5 VO Voice, < 10 ms latency 6 IC Internetwork Control 7 NC Network Control
IEEE 802.1Q-2005
QoS Tools Review: Classification & Marking Tools Layer 2 Marking: IEEE 802.1Q/p CoS
class-map VOICE match cos 5
policy-map MARK-COS class VIDEO set cos 4
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
§ IP Precedence (relegated): Three most significant bits of ToS byte are called IP Precedence (IPP)—other bits unused
§ Differentiated Services: Six most significant bits of ToS byte are called DiffServ Code Point (DSCP)—remaining two bits used for Explicit Congestion Notification (ECN)
§ DSCP and ECN are also used in IPv6
7 6 5 4 3 2 1 0
ID Offset TTL Protocol FCS IP SA IP DA Data Length Version/ Header_Len
ToS Byte
DiffServ Code Point (DSCP) IP ECN
IPv4 Packet
IP Precedence Unused
QoS Tools Review: Classification & Marking Tools Layer 3 Marking: IP Type of Service (ToS) Byte
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
x x x y y 0 AFxy Class Drop
Precedence
DSCP
IP Header ToS Byte
EF
Default Forwarding (Best Effort)
RFC 2474
AF11
AF21
AF31
AF41
AF12 AF13
AF22 AF23
AF32 AF33
AF42 AF43
Expedited Forwarding RFC 3246
Assured Forwarding RFC 2597
Per-Hop Behaviors (PHB) DiffServ Code Points (DSCP)
101110
001010 001100 001110
010010 010100 010110
011010 011100 011110
100010 100100 100110
Class 1
Class 2
Class 3
Class 4
Low Drop Pref
Med Drop Pref
High Drop Pref
000000
46
10 12 14
18 20 22
26 28 30
34 36 38
0 DF
QoS Tools Review: Classification & Marking Tools Layer 3 Marking: DSCP Per-Hop Behaviors (PHBs)
class-map VOICE match dscp ef
policy-map SET-DSCP class VIDEO set dscp af41
Scavenger (Per Domain
Behavior—RFC3662)
001000 8
CS1
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
Label TTL
MPLS Shim Header
EXP S
Payload
Frame Encapsulation
3 2 1 0
MPLS EXP S
Layer-2 Header
QoS Tools Review: Classification & Marking Tools Layer 2.5 Marking: MPLS Experimental (EXP) Bits
class-map VOICE match mpls experimental 5
policy-map SET-MPLS-EXP class VIDEO set mpls experimental imposition 4 <or> set mpls experimental topmost 4
Label Header Label Header
Label Stack
S=Bottom of Stack bit MPLS Experimental bits
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
ToS IP SA IP DA Src Port
Dst Port
Protocol
§ Identifies over 1000 applications and protocols
§ Application payload deep packet inspection
§ Supports application media-sub-component classification
TCP/UDP Segment
Deep Packet Inspection
Data Payload IP Packet
QoS Tools Review: Classification & Marking Tools Layer 7 Classification: Network Based Application Recognition (NBAR/NBAR2)
class-map CISCO-JABBER-VOICE match protocol cisco-jabber-audio class-map CISCO-JABBER-VIDEO match protocol cisco-jabber-video class-map CISCO-JABBER-MESSAGING match protocol cisco-jabber-im class-map CISCO-JABBER-SIGNALING match protocol cisco-jabber-control
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Tools Review: Policing & Shaping Tools
§ Policers: – perform checks for traffic violations against a configured rate and take immediate
prescribed actions (such as remarking or dropping) – policers do not delay traffic – policers may be applied to the data plane or the control plane
§ Shapers: – smooth out traffic flows so that it never exceeds the configured rate – if the offered traffic momentarily spikes above the contracted rate, the excess traffic is
buffered and delayed until the offered traffic once again dips below the defined rate – shapers usually are employed to meet a Service Level Agreement (SLA)
Policers vs. Shapers
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Action Action
Overflow
B<Tc B<Te
Conform Exceed Violate
CBS EBS
CIR
Yes Yes
No No
Action
Packet of Size B
QoS Tools Review: Policing & Shaping Tools RFC 2697 Single-Rate Three-Color Marker
policy-map RFC2697-POLICER class CLASS-1 police cir 500000 bc 10000 be 10000 conform-action set-dscp-transmit af31 exceed-action set-dscp-transmit af32 violate-action set-dscp-transmit af33
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Action Action
B>Tp B>Tc
Exceed Violate
PBS CBS
PIR
Yes Yes
No No
Conform
Action
Packet of Size B
CIR
QoS Tools Review: Policing & Shaping Tools RFC 2698 Two-Rate Three-Color Marker
policy-map RFC2698-POLICER class CLASS-2 police cir 500000 bc 10000 pir 100000 be 10000 conform-action set-dscp-transmit af31 exceed-action set-dscp-transmit af32 violate-action set-dscp-transmit af33
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Tools Review: Policing & Shaping Tools Shaping Effect on Traffic Patterns
With Traffic Shaping
Without Traffic Shaping Line Rate
Shaped Rate
Traffic Shaping Limits the Transmit Rate to a Value Lower Than Line Rate
policy-map CLASS-BASED-SHAPER class class-default shape average 10 Mbps <or> shape peak 10 Mbps
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Packets In Packets Out Tx-Ring
IOS Interface Buffers
If the Tx-Ring is filled to capacity, then the IOS software knows that the interface
is congested and it should activate any LLQ/CBWFQ policies that have been
applied to the interface
QoS Tools Review: Queuing & Dropping Tools Tx-Ring
interface Serial2/0 tx-ring-limit 4
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Packets In
Packets Out
Fair-Queuing Sorter/Pre-Sorter
A flow is defined by five matching tuples: Source Address + Source Port
Destination Address + Destination Port Layer 4 Protocol (TCP or UDP)
QoS Tools Review: Queuing & Dropping Tools (Flow-Based) Fair-Queuing
policy-map FQ class class-default fair-queue
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Packets In
Packets Out
IOS Interface Buffers
Tx-Ring
Multimedia Conferencing CBWFQ
Multimedia Streaming CBWFQ
Network Control CBWFQ
Call Signaling CBWFQ
OAM CBWFQ
Transactional Data CBWFQ
Bulk Data CBWFQ
Best Effort / Default CBWFQ
Scavenger CBWFQ
CBWFQ Scheduler
FQ
FQ
FQ
FQ Pre-Sorters
FQ
FQ
QoS Tools Review: Queuing & Dropping Tools CBWFQ policy-map CBWFQ
class NETWORK-CONTROL bandwidth percent 5 class SIGNALING bandwidth percent 5 class OAM bandwidth percent 5 class MM-CONFERENCING bandwidth percent 10 fair-queue
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Packets In Packets Out
IOS Interface Buffers
Tx-Ring CBWFQ
Scheduler
LLQ
1 Mbps VOICE Policer
FQ Pre-Sorters
CBWFQs
policy-map LLQ class VOICE priority 1000
QoS Tools Review: Queuing & Dropping Tools LLQ: Single-LLQ Operation and Configuration
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Packets In Packets Out
Tx-Ring CBWFQ
Scheduler
LLQ
1 Mbps VOICE Policer
4 Mbps Bscst-Video
Policer
5 Mbps RT-Interactive
Policer
CBWFQs
policy-map MULTI-LLQ class VOICE priority 1000 class BROADCAST-VIDEO priority 4000 class REALTIME-INTERACTIVE priority 5000
QoS Tools Review: Queuing & Dropping Tools LLQ: Multi-LLQ Operation and Configuration
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Time
Bandwidth Utilization 100%
BW
Tail Drop Three Traffic Flows Start at Different Times
Another Traffic Flow Starts at This Point
QoS Tools Review: Queuing & Dropping Tools The Need for Congestion Avoidance
§ all TCP flows synchronize in waves § TCP synchronization wastes available bandwidth
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Bulk Data CBWFQ Fair- Queuing Pre-Sorter
AF13 Minimum WRED Threshold: Begin randomly dropping AF13 Packets
AF12 Minimum WRED Threshold: Begin randomly dropping AF12 Packets
AF11 Minimum WRED Threshold: Begin randomly dropping AF11 Packets
Maximum WRED Thresholds for AF11, AF12 and AF13 are set to the tail of the queue in this example
Front of
Queue
Tail of
Queue Direction
of Packet Flow
policy-map BULK-WRED class BULK bandwidth percent 10 random-detect dscp-based
QoS Tools Review: Queuing & Dropping Tools DSCP-Based WRED
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
RFC 3168: IP Explicit Congestion Notification
CE Bit: Congestion Experienced
ECT Bit: ECN-Capable Transport
ID Offset TTL Proto FCS IP SA IP DA Data Len Version Length
ToS Byte
IPv4 Packet *However IP ECN is also supported on IPv6
7 6 5 4 3 2 1 0 DiffServ Code Point (DSCP) CE ECT
policy-map ECN class class-default bandwidth percent 25 random-detect ecn
QoS Tools Review: Queuing & Dropping Tools IP Explicit Congestion Notification (IP ECN)
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
A Case Study in Strategic QoS Design
§ QoS Tools Review
§ QoS Design Best Practices
§ QoS Design Strategies
§ QoS Design Case Study
Agenda
29
QoS Design Best Practices
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Design Best Practices
§ Always enable QoS policies in hardware—rather than software—whenever a choice exists
§ Classify and mark applications as close to their sources as technically and administratively feasible
§ Use DSCP marking whenever possible
§ Follow standards-based DSCP PHB markings to ensure interoperability and future expansion
Classification & Marking Best Practices
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Design Best Practices
§ Police traffic flows as close to their source as possible
§ Whenever possible, markdown according to standards-based rules
§ For Example: Assured Forwarding Traffic (AF21 example) § Conforming AF21 traffic is marked/remarked AF21 § Exceeding AF21 traffic is remarked AF22 § Violating AF21 traffic is remarked AF23
Policing and Remarking Best Practices
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Design Best Practices
§ Enable queuing policies at every node that has the potential for congestion
§ Whenever possible, assign each application class to its own dedicated queue
§ Use only platforms and/or service providers that offer a minimum of four standards-based queuing behaviors: – An RFC 3246 Expedited Forwarding Per-Hop Behavior – An RFC 2597 Assured Forwarding Per-Hop Behavior – An RFC 2474 Default Forwarding Per-Hop Behavior – An RFC 3662 Lower Effort Per-Domain Behavior
Queuing Best Practices
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Design Best Practices
§ Enable DSCP-based WRED on AF queues
§ Enable WRED on the DF queue
§ Do not enable DSCP-based WRED on the EF queue
§ Do not enable WRED on control traffic application class queues
§ WRED is not required on the Scavenger queue
§ Optional: Tune WRED thresholds consistently—for example: § Set the minimum WRED thresholds for AFx3 to 60% of the queue depth § Set the minimum WRED thresholds for AFx2 to 70% of the queue depth § Set the minimum WRED thresholds for AFx1 to 80% of the queue depth § Set all maximum WRED thresholds to 100%
WRED Principles
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
• EF Queue Recommendations: • Limit the amount of strict priority queuing to 33% of link bandwidth capacity • Govern strict-priority traffic with an admission control mechanism • Do not enable WRED on this queue
• AF Queue Recommendations: • Provision guaranteed bandwidth allocations according to application requirements • Enable DSCP-based WRED on this queue(s)
• DF Queue Recommendations: • Provision at least 25 percent of link bandwidth for the default Best Effort class • Enable WRED (effectively RED) on the default class
• Scavenger Queue Recommendations: • Assign minimum bandwidth to the Scavenger-class queue • WRED is not required on the Scavenger-class queue
QoS Design Best Practices Per-Hop Behavior Principles
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
A Case Study in Strategic QoS Design
§ QoS Tools Review
§ QoS Design Best Practices
§ QoS Design Strategies
§ QoS Design Case Study
Agenda
36
QoS Design Strategies
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Design Strategies
§ By 2017… – Global IP traffic will triple – The number of IP devices will be more than triple the global population – Wireless traffic will exceed wired – Half of all IP traffic will be non-PC traffic (smartphones, tablets etc.) – 80-90% of consumer internet traffic will be video
Macro Trends in Internet Usage
38
http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/white_paper_c11-481360_ns827_Networking_Solutions_White_Paper.html
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Design Strategies RFC 4594: Configuration Guidelines for DiffServ Service Classes
39
Application Class
Per-Hop Behavior
Admission Control
Queuing & Dropping
Application Examples
VoIP Telephony EF Required Priority Queue (PQ) Cisco IP Phones (G.711, G.729)
Broadcast Video CS5 Required (Optional) PQ Cisco IP Video Surveillance / Cisco Enterprise TV
Realtime Interactive CS4 Required (Optional) PQ Cisco TelePresence
Multimedia Conferencing AF4 Required BW Queue + DSCP WRED Cisco Jabber, Cisco WebEx
Multimedia Streaming AF3 Recommended BW Queue + DSCP WRED Cisco Digital Media System (VoDs)
Network Control CS6 BW Queue EIGRP, OSPF, BGP, HSRP, IKE
Signaling CS3 BW Queue SCCP, SIP, H.323
Ops / Admin / Mgmt (OAM) CS2 BW Queue SNMP, SSH, Syslog
Transactional Data AF2 BW Queue + DSCP WRED ERP Apps, CRM Apps, Database Apps
Bulk Data AF1 BW Queue + DSCP WRED E-mail, FTP, Backup Apps, Content Distribution
Best Effort DF Default Queue + RED Default Class
Scavenger CS1 Min BW Queue (Deferential) YouTube, iTunes, BitTorent, Xbox Live
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
QoS Design Strategies Business Requirements Will Continue to Evolve and Expand over Time
40
Time
Critical Data
Realtime
4-Class Model
Best Effort
Signaling / Control Call Signaling
Critical Data
Interactive Video
Voice 8-Class Model
Scavenger
Best Effort
Streaming Video
Network Control Network Management
Realtime Interactive
Transactional Data
Multimedia Conferencing
Voice 12-Class Model
Bulk Data
Scavenger Best Effort
Multimedia Streaming
Network Control
Broadcast Video
Call Signaling
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Transactional Data
Realtime
4-Class Model
Best Effort
Control
AF21
EF
DSCP
DF
CS3
Real%me 33%
Control 7%
Best Effort 25%
Transactional Data 35%
QoS Design Strategies Basic (4-Class) QoS Strategy
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Signaling
Transactional Data
Interactive Video
Voice 8-Class Model
Scavenger Best Effort
Streaming Video
Network Control CS3
AF2
AF41
EF DSCP
CS1 DF
AF31
CS6
Interac%ve Video 23%
Network Control 5%
Signaling 2%
Transac%onal Data 24%
Best Effort 25%
Voice 10%
Scavenger 1%
Streaming Video 10%
QoS Design Strategies Intermediate (8-Class) QoS Strategy
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Voice 10%
Streaming Video 10%
OAM
Realtime Interactive
Transactional Data
Multimedia Conferencing
Voice 12-Class Model
Bulk Data
Scavenger Best Effort
Multimedia Streaming Network Control
Broadcast Video
Signaling CS2
CS4
AF2
AF4
EF DSCP
AF1
CS1 DF
AF3 CS6
CS5
CS3
QoS Design Strategies Advanced (12-Class) QoS Strategy
Broadcast Video 10%
Real%me Interac%ve 13%
Network Control 2%
OAM 3%
Signaling 2%
Mul%media Streaming
10%
Mul%media Conferencing
10%
Transac%onal Data 10%
Bulk Data 4%
Scavenger 1%
Best Effort 25%
Voice 10%
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Voice
Interactive Video
Broadcast Video
Realtime Interactive
Transactional Data
Network Control
Network Management (OAM)
Signaling Multimedia Streaming Multimedia
Conferencing
Bulk Data
Scavenger
Best Effort
4-Class Model
8-Class Model
12-Class Model
Strategic Bandwidth Allocations Consistency in Bandwidth Allocation as QoS Models Evolve
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
A Case Study in Strategic QoS Design
§ QoS Tools Review
§ QoS Design Best Practices
§ QoS Design Strategies
§ QoS Design Case Study
Agenda
45
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
“Mission-Critical Data”
Voice 4-Class Model
Best Effort
Signaling
AF31
EF DSCP
DF
CS3
Voice 33%
Signaling 7%
“Mission-‐Cri%cal Data” 35%
Best Effort 25%
Case Study: Tifosi Software Original Four-Class QoS and Queuing Models
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Case Study: Tifosi Software
§ The imminent deployment of two dozen Cisco TelePresence Systems – one each in every major engineering and sales office – with future plans for more to come
§ The emerging popularity of multimedia collaboration applications – including Cisco WebEx, Cisco Jabber, as well as Microsoft Lync
§ Too many applications classified as “mission critical” – including email, file-transfers, backup operations, etc.
§ The identification of “substantial” amounts of non-business traffic on the network during work hours – including YouTube, BitTorrent and iTunes downloads, as well as gaming traffic
§ A desire for greater overall QoS policy-consistency
Current Business Requirements
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Voice 10%
Realtime Interactive 23%
Signaling 2%
Multimedia Conferencing
10%
Transactional Data 25%
Bulk Data 4%
Scavenger 1%
Best Effort 25%
Signaling
“Mission Critical” Data Multimedia Conferencing
Voice 8-Class Model
Scavenger Best Effort
Bulk Data
Realtime Interactive CS3
AF21 AF41
EF DSCP
CS1 DF
AF11
CS4
Case Study: Tifosi Software Proposed Eight-Class Strategic QoS Model
Transactional Data
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Signaling
Transactional Data
Realtime Interactive
Voice
Application Classes
Scavenger
Best Effort
Multimedia Conferencing
CS3
AF2
CS4
EF
DSCP
CS1
DF
AF4
Bulk Data AF1
Q1 Priority Queue
Queue 4 (5%)
Queue 2 (30%)
Default Queue Queue 3 (35%)
Q2T2
Q4T2 Q4T1
Q2T1
EF CS4
CS3
DF
AF1 CS1
AF4
AF2
1P3Q3T
Case Study: Tifosi Software Campus Access Catalyst 3750 Eight-Class (1P3Q3T) Egress Queuing Model
1P3Q3T= 1 Priority Queue
3 (Non-Priority)
Queues (each with)
3 Drop Thresholds Q2T1= Queue 2, Threshold 1
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
1P7Q1T (+DBL)
Signaling
Transactional Data
Realtime Interactive
Voice
Application Classes
Scavenger
Best Effort
Multimedia Conferencing
CS3
AF2
CS4
EF
DSCP
CS1
DF
AF4
Signaling Queue (2% BWR)
Transactional Data Queue
(25% BWR + DBL)
Priority Queue
Scavenger (1% BWR)
Default Queue (25% BWR + DBL)
Multimedia Conferencing Queue
(10% BWR + DBL)
EF
DF
CS1
AF2
AF4
CS4
CS3
Bulk Data AF1 Bulk Data Queue
(4% BWR + DBL) AF1
Case Study: Tifosi Software Campus Distribution Catalyst 4500 Eight-Class (1P7Q1T+DBL) Queuing Model
DBL= Dynamic Buffer Limiting BWR= Bandwidth Remaining
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
8Q4T/1P7Q4T
Signaling
Transactional Data
Realtime Interactive
Voice
Application Classes
Scavenger
Best Effort
CS3
AF2
CS4
EF
DSCP
CS1
DF
Signaling Queue (2% BW/BWR)
Transactional Data Queue (25% BW/BWR
+ DSCP-based WRED)
Realtime-Queue (33% BW/Priority)
Scavenger Queue (1% BW/BWR) Default Queue
(25% BW/BWR + WRED)
EF
DF
CS1
AF2
CS3
CS4
Multimedia Conferencing AF4 Multimedia Conferencing Queue
(10% BW /BWR + DSCP-based WRED) AF4
Bulk Data AF1 Bulk Data Queue
(4% BW/BWR + DSCP-based WRED) AF1
Case Study: Tifosi Software Campus Core Catalyst 6500 Eight-Class (8Q4T & 1P7Q4T) Queuing Models
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Signaling
Transactional Data
Realtime Interactive
Voice
Application Classes
Scavenger
Best Effort
CS3àDSCP 33
AF2
CS4
EF
DSCP
CS1
DF
Multimedia Conferencing AF4
Bulk Data AF1
Case Study: Tifosi Software Centralized Cisco 5508 Wireless LAN Controller 802.11e/WMM Queuing Model
Platinum
802.11e User Priorities (UP) & WLC Access Categories
Silver
Gold
Bronze
UP 7
UP 5
UP 3
UP 2
UP 6
UP 4
UP 0
UP 1
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Transactional Data
Video / Realtime Interactive / Signaling
Bulk Data
Best Effort
Voice
Application Class
Internetwork Control
Network Control
AF2
AF4 / CS4 / CS3*
AF1
DF
EF
DSCP
CS6
N/A
8Q2T / 1P7Q4T Q1/PQ
8Q2T: 10% BW + 10% QL 1P7Q4T: Priority + 10% QL
CoS 5
Q2 8Q2T: 5% BW + 5% QL
1P7Q4T: 5% BWR + 5% QL
CoS 7
Q3 8Q2T: 5% BW + 5% QL
1P7Q4T: 5% BWR + 5% QL CoS 6
Q4 8Q2T: 35% BW + 35% QL + WRED
1P7Q4T: 35% BWR + 35% QL + WRED
CoS 4
Q6 8Q2T: 10% BW + 10% QL + WRED
1P7Q4T: 10% BWR + 10% QL + WRED
CoS 2
Q7 8Q2T: 10% BW + 10% QL + WRED
1P7Q4T: 10% BWR + 10% QL + WRED CoS 1
Q-Default 8Q2T: 25% BW + 25% QL + WRED
1P7Q4T: 35% BWR + 25% QL + WRED CoS 0
CoS 2
CoS 4
CoS 1
CoS 0
CoS 5
CoS
CoS 6
CoS 7
Q5 8Q2T: 0% BW + 0% QL
1P7Q4T: 0% BWR + 0% QL
CoS 3
Case Study: Tifosi Software DC Core Cisco Nexus 7000 M2 (8Q2T Ingress / 1P7Q4T Egress) Queuing Model
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Q-Default (In-Q2 / Out-Q4)
4Q1T: 25% BW + 32% QL 1P3Q1T: 40% BWR
FCoE
Transactional Data
Video / Signaling
Bulk Data / vMotion
Best Effort
Voice / Realtime Interactive
Application Class
Internetwork Control
Network Control
N/A
AF2
AF4 / CS3*
AF1 / N/A
DF
EF / CS4
DSCP
CS6
4Q1T / 1P3Q1T
In-Q1 / Out-PQ1 4Q1T: 25% BW + 7% QL 1P3Q1T: Priority Level 1
CoS 7 CoS 6 CoS 5
In-Q3 / Out-Q3 4Q1T: 25% BW + 31% QL
1P3Q1T: 20% BWR
CoS 4
In-Q4 / Out-Q2 No Drop
4Q1T: 25% BW + 30% QL 1P3Q1T: 40% BWR
CoS 3
CoS 2
CoS 1
CoS 0
CoS 3
CoS 2
CoS 4
CoS 1*
CoS 0
CoS 5
CoS
CoS 6
CoS 7
Case Study: Tifosi Software DC Fabric Cisco Nexus 7000 F2 (nq-7e: 4Q1T / 1P3Q1T) Queuing Model
N/A
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Signaling
Transactional Data
Multimedia Conferencing
Voice
Customer 8-Class Model
Scavenger
Best Effort
Bulk Data
CS3 è CS4
AF21
AF41è AF31
EF
DSCP
CS1
DF
AF11
Realtime Interactive CS4
SP Six-Class Model Classes-of-Service
SP-REALTIME-CLASS (RTP) LLQ 10%
EF CS5
SP-AF4-CLASS (RTP) CBWFQ 25% BW +
DSCP-WRED AF41 CS4
SP-AF1-CLASS (Control) CBWFQ 5% BW +
DSCP-WRED
SP-DEFAULT-CLASS CBWFQ 25% BW +
WRED
DF
SP-AF3-CLASS (UDP) CBWFQ 10% BW +
DSCP-WRED AF31 CS3
SP-AF2-CLASS (TCP) CBWFQ 25% BW +
DSCP-WRED
AF21 CS2
AF11 CS1
Case Study: Tifosi Software MPLS VPN Customer-Edge Enterprise-to-Service Provider Mapping
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Recommended Reading
§ Release Date: Jan 2014 § Comprehensive QoS design
guidance for PINs and platforms: • Campus Catalyst 3750/4500/6500 • WLAN WLC 5508 / Catalyst 3850 NGWC • Data Center Nexus 1000V/2000/5500/7000 • WAN & Branch Cisco ASR 1000 / ISR G2 • MPLS VPN Cisco ASR 9000 / CRS-3 • IPSec VPNs Cisco ISR G2
• ISBN: 1-58714-369-0
56
Ken Briley
http://www.ciscopress.com/store/end-to-end-qos-network-design-quality-of-service-for-9781587143694
© 2014 Cisco and/or its affiliates. All rights reserved. Session ID CP-1002 Cisco Public
Thank You for Joining Us Today
57
Download a copy of todays slides using the link in the chat.
Today’s webcast will be available on-demand within 48hrs.
Please complete the survey after closing the WebEx event.