Quality of Service in the Internet Dr Vasil

  • Slides: 45
Download presentation
Quality of Service in the Internet Dr. Vasil Y. Hnatyshin 11/3/2020 Rowan University Vasil

Quality of Service in the Internet Dr. Vasil Y. Hnatyshin 11/3/2020 Rowan University Vasil Y. Hnatyshin 1

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated Services Differentiated Services Bandwidth Distribution Scheme Summary and Conclusions 11/3/2020 Rowan University Vasil Y. Hnatyshin 2

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated Services Differentiated Services Bandwidth Distribution Scheme Summary and Conclusions 11/3/2020 Rowan University Vasil Y. Hnatyshin 3

What is Qo. S? What is Quality of Service? What parameters influences perceived network

What is Qo. S? What is Quality of Service? What parameters influences perceived network Qo. S? Why do we need Qo. S in the Internet? 11/3/2020 Rowan University Vasil Y. Hnatyshin 4

Internet Architecture What architectural model is being used in today’s Internet? What are IP,

Internet Architecture What architectural model is being used in today’s Internet? What are IP, TCP, and UDP and how do they work? 11/3/2020 Rowan University APPLICATION TCP/UDP IP MAC TCP/IP Internet Architecture Vasil Y. Hnatyshin 5

How does Internet work? Widener Rowan APPLICATION INTERNET TCP/UDP IP APPLICATION IP MAC TCP/UDP

How does Internet work? Widener Rowan APPLICATION INTERNET TCP/UDP IP APPLICATION IP MAC TCP/UDP MAC IP IP MAC ISP 11/3/2020 ISP Rowan University Vasil Y. Hnatyshin 6

How to achieve Qo. S? Provisioning Traffic differentiation n 11/3/2020 In ISP routers (and

How to achieve Qo. S? Provisioning Traffic differentiation n 11/3/2020 In ISP routers (and the end nodes) In the Internet core routers Combination of the above Rowan University Vasil Y. Hnatyshin 7

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated Services Differentiated Services Bandwidth Distribution Scheme Summary and Conclusions 11/3/2020 Rowan University Vasil Y. Hnatyshin 8

Qo. S Building Blocks Incoming Packet Classifier Rate Controller Admission Control Scheduler Outgoing Packet

Qo. S Building Blocks Incoming Packet Classifier Rate Controller Admission Control Scheduler Outgoing Packet 11/3/2020 Rowan University Vasil Y. Hnatyshin 9

Classifiers Application block Source/Destination IP addresses Source/Destination ports Transport protocol Type of Service Field

Classifiers Application block Source/Destination IP addresses Source/Destination ports Transport protocol Type of Service Field (Differentiated Services) Version Header Type of Service Length Total Packet Length … TTL Protocol Header Checksum Source Address Destination Address Options (Variable Size) 11/3/2020 Rowan University Vasil Y. Hnatyshin 10

Classifiers Class #1 Classifier Class #2 11/3/2020 Rowan University Vasil Y. Hnatyshin 11

Classifiers Class #1 Classifier Class #2 11/3/2020 Rowan University Vasil Y. Hnatyshin 11

Admission control Ensures that the load on a particular link is manageable by determining

Admission control Ensures that the load on a particular link is manageable by determining if the new flow can be admitted into the network. Admission Link utilization is 60 % Link utilization is 90 % Control 11/3/2020 Rowan University Vasil Y. Hnatyshin 12

Rate Controllers What are rate controllers and how rate control is achieved? n n

Rate Controllers What are rate controllers and how rate control is achieved? n n Shapers Droppers Common rate control mechanisms n n 11/3/2020 Token bucket Timestamp driven Rowan University Vasil Y. Hnatyshin 13

Rate Controllers Processing Queue Rate Controlle r Delay Queue Dropper 11/3/2020 Rowan University Vasil

Rate Controllers Processing Queue Rate Controlle r Delay Queue Dropper 11/3/2020 Rowan University Vasil Y. Hnatyshin 14

Schedulers What are schedulers? Scheduling mechanisms: n n FIFO Priority Queuing w Potential problem?

Schedulers What are schedulers? Scheduling mechanisms: n n FIFO Priority Queuing w Potential problem? w Preemptive vs. non-preemptive n 11/3/2020 Weighted-Fair Queuing Rowan University Vasil Y. Hnatyshin 15

Priority Scheduler Priority Queue #1 Outgoing Link Priority Queue #2 Priority Queue #3 11/3/2020

Priority Scheduler Priority Queue #1 Outgoing Link Priority Queue #2 Priority Queue #3 11/3/2020 Rowan University Vasil Y. Hnatyshin 16

Qo. S Blocks Revisited Incoming Packet Classifier Admission Control SLA Rate Controller Scheduler Outgoing

Qo. S Blocks Revisited Incoming Packet Classifier Admission Control SLA Rate Controller Scheduler Outgoing Packet 11/3/2020 Rowan University Vasil Y. Hnatyshin 17

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated Services Differentiated Services Bandwidth Distribution Scheme Summary and Conclusions 11/3/2020 Rowan University Vasil Y. Hnatyshin 18

Integrated Services Simplified Idea: for each newly admitted flow reserve network resources at each

Integrated Services Simplified Idea: for each newly admitted flow reserve network resources at each node on the flow’s path from source to destination. Uses per-flow resource reservation protocol called resource reservation protocol (RSVP). 11/3/2020 Rowan University Vasil Y. Hnatyshin 19

Int. Serv Example Internet Widener Core 1 Core 3 ISP 1 ISP 2 ISP

Int. Serv Example Internet Widener Core 1 Core 3 ISP 1 ISP 2 ISP 4 Core 2 Rowan FTP connection from Rowan to CNN Video conference call from Rowan to CNN Allocated resources for corresponding connections FTP connection from Widener to CNN Video conference call from Widener to CNN 11/3/2020 Rowan University Vasil Y. Hnatyshin 20

Int. Serv Summary Main Advantage • Support for fine-grained perflow Qo. S 11/3/2020 Rowan

Int. Serv Summary Main Advantage • Support for fine-grained perflow Qo. S 11/3/2020 Rowan University Main Disadvantage • Resource hogging • Not Scalable Vasil Y. Hnatyshin 21

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated Services Differentiated Services Bandwidth Distribution Scheme Summary and Conclusions 11/3/2020 Rowan University Vasil Y. Hnatyshin 22

Differentiated Services Goal: Provide scalable Qo. S. Diff. Serv Implementation: n n 11/3/2020 Establishes

Differentiated Services Goal: Provide scalable Qo. S. Diff. Serv Implementation: n n 11/3/2020 Establishes a few classes/aggregates Classifies arriving traffic into one of predefined classes. Packets are differentiated based on the DSCP marking set in the IP header. Core routers treat arriving traffic based on the packet’s class. Rowan University Vasil Y. Hnatyshin 23

Diff. Serv Example Internet Widener Core 1 Core 3 ISP 1 ISP 2 ISP

Diff. Serv Example Internet Widener Core 1 Core 3 ISP 1 ISP 2 ISP 4 Core 2 Rowan FTP connection from Rowan to CNN Video conference call from Rowan to CNN DS Class A DS Class B FTP connection from Widener to CNN Video conference call from Widener to CNN 11/3/2020 Rowan University Vasil Y. Hnatyshin 24

Diff. Serv Summary Main Advantage • Scalability 11/3/2020 Rowan University Main Disadvantage • Providing

Diff. Serv Summary Main Advantage • Scalability 11/3/2020 Rowan University Main Disadvantage • Providing only coarse-grained peraggregated Qo. S Vasil Y. Hnatyshin 25

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated Services Differentiated Services Bandwidth Distribution Scheme (BDS) Summary and Conclusions 11/3/2020 Rowan University Vasil Y. Hnatyshin 26

Idea of the BDS Flow request certain amounts of bandwidth upon activation. Bandwidth flow

Idea of the BDS Flow request certain amounts of bandwidth upon activation. Bandwidth flow requirements are aggregated and distributed in the network via Resource Distribution and Feedback Protocol (RDF Protocol). Based on the aggregate flow requirements individual edge routers control the rate of the traffic. 11/3/2020 Rowan University Vasil Y. Hnatyshin 27

BDS Example Internet Widener Core 1 Core 3 ISP 1 ISP 2 ISP 4

BDS Example Internet Widener Core 1 Core 3 ISP 1 ISP 2 ISP 4 Core 2 Rowan FTP connection from Rowan to CNN Video conference call from Rowan to CNN Aggregate Flow Requirements FTP connection from Widener to CNN Video conference call from Widener to CNN 11/3/2020 Rowan University Vasil Y. Hnatyshin 28

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated

Outline Introduction Qo. S via traffic differentiation n n Qo. S Building Blocks Integrated Services Differentiated Services Bandwidth Distribution Scheme Summary and Conclusions 11/3/2020 Rowan University Vasil Y. Hnatyshin 29

Summary and Conclusions The BDS approach has potential to become superior to current Int.

Summary and Conclusions The BDS approach has potential to become superior to current Int. Serv and Diff. Serv approaches. BDS Advantages: n n Scalability Per-flow Qo. S Fairness Congestion control 11/3/2020 Rowan University BDS Disadvantages: n n Not tested under reallife scenarios Overhead of the RDF protocol Vasil Y. Hnatyshin 30

Ongoing BDS Research Evaluation of the BDS approach under more realistic scenarios. Dealing with

Ongoing BDS Research Evaluation of the BDS approach under more realistic scenarios. Dealing with the inter-domain traffic. Extending the BDS framework to mobile environment. Influence of the BDS model on TCP traffic. 11/3/2020 Rowan University Vasil Y. Hnatyshin 31

The BDS Publications V. Hnatyshin and A. S. Sethi, “Reducing load distribution overhead with

The BDS Publications V. Hnatyshin and A. S. Sethi, “Reducing load distribution overhead with message aggregation, ” Proc of IPCCC’ 03, Phoenix, AZ (April 2003) V. Hnatyshin and A. S. Sethi, “Fair and Scalable Load Distribution in the Internet, ” Proc. 3 rd International Conference on Internet Computing, Las Vegas, NV (June 2002). V. Hnatyshin and A. S. Sethi, “Achieving Fair and Predictable Service Differentiation Through Traffic Degradation Policies, ” Proc. SPIE Qo. S 2001, Conference on Quality of Service over Next-Generation Data Networks, Denver, CO (Aug. 2001). V. Hnatyshin and A. S. Sethi, “Avoiding Congestion Through Dynamic Load Control, ” Proc. ITCom-2001, SPIE's International Symposium on The Convergence of Information Technologies and Communications, Denver, CO (Aug. 2001). 11/3/2020 Rowan University Vasil Y. Hnatyshin 32

The End! Rowan University Y. Hnatyshin Vasil

The End! Rowan University Y. Hnatyshin Vasil

Network Architecture Network Domain Core 1 The Internet Network Domain Core 2 Edge B

Network Architecture Network Domain Core 1 The Internet Network Domain Core 2 Edge B Edge A Edge C Core 3 Edge D Network Domain 11/3/2020 Rowan University Vasil Y. Hnatyshin 34

Diff. Serv Scalability Edge Routers: Network Domain Core 1 Edge B Edge A Edge

Diff. Serv Scalability Edge Routers: Network Domain Core 1 Edge B Edge A Edge C 11/3/2020 Core 2 Core 3 Edge D Rowan University • Maintain all per-flow information via SLA • Perform traffic classification • Mark arriving packets Core Routers: • Maintain only class information • Treat arriving traffic based on the DSCP marking Vasil Y. Hnatyshin 35

BDS Edge Node Data Structures Flow List Path Table … Source Destination Data …

BDS Edge Node Data Structures Flow List Path Table … Source Destination Data … … Egress node SLA Table Source … … Data Destination … Link Table Complete Path … Link Capacity IP Address Data Arrival Rate IP Address Data Aggregate RBR IP Address Data Path List … … 11/3/2020 Rowan University Vasil Y. Hnatyshin 36

BDS Core Node Data Structures Interfaces Table IP Address Data … 11/3/2020 Link Capacity

BDS Core Node Data Structures Interfaces Table IP Address Data … 11/3/2020 Link Capacity Arrival Rate Aggregate RBR Edge Nodes List … Rowan University Vasil Y. Hnatyshin 37

BDS Objectives Scalability Per-flow Qo. S Fair distribution of bandwidth Congestion Control 11/3/2020 Rowan

BDS Objectives Scalability Per-flow Qo. S Fair distribution of bandwidth Congestion Control 11/3/2020 Rowan University Vasil Y. Hnatyshin 38

BDS Scalability Edge Routers: Network Domain Core 1 Edge B Edge A Edge C

BDS Scalability Edge Routers: Network Domain Core 1 Edge B Edge A Edge C 11/3/2020 Core 2 Core 3 Edge D Rowan University • Maintain all per-flow information via SLA • Perform traffic classification • Adjust transmission rate of the flows based on the network feedback Core Routers: • Maintain only aggregate flow requirements • Provide network feedback to the edge routers Vasil Y. Hnatyshin 39

BDS Per-flow Qo. S Admission control Resource management Control protocol Note: Fairness and congestion

BDS Per-flow Qo. S Admission control Resource management Control protocol Note: Fairness and congestion control achieved via same modules 11/3/2020 Rowan University Vasil Y. Hnatyshin 40

Admission Control Flow Requirements: Requested Bandwidth Range (RBR) [Min Rate, Max Rate] = Aggregate

Admission Control Flow Requirements: Requested Bandwidth Range (RBR) [Min Rate, Max Rate] = Aggregate RBR Admission control test: -- A new flow -- A path of a flow -- A flow 11/3/2020 Rowan University -- A link -- Capacity of link k -- A set of flows that travel through link k Vasil Y. Hnatyshin 41

Resource Management Definitions of Fairness: n Proportional Fairness n Maximizing Utility Fairness Per-Flow Resource

Resource Management Definitions of Fairness: n Proportional Fairness n Maximizing Utility Fairness Per-Flow Resource Allocation 11/3/2020 Rowan University Vasil Y. Hnatyshin 42

The RDF Protocol Purpose: n Distribute the Aggregate RBR among the nodes in the

The RDF Protocol Purpose: n Distribute the Aggregate RBR among the nodes in the network. Path Probing Phase: n Edge routers periodically probe the network to discover the route changes (e. g. aggregate RBR, excess bandwidth). RBR Update Phase: n Edge nodes notify the core routers about the change of the aggregate RBR due to flow activation or termination. Notification Phase: n Core routers notify the edge nodes about congestion. 11/3/2020 Rowan University Vasil Y. Hnatyshin 43

The RDF Protocol Example Flow F 2 requests to enter the network at Edge

The RDF Protocol Example Flow F 2 requests to enter the network at Edge 2 to travel to Edge 5 F 1 Edge 1 CN 1. Initiates the Path Probing Phase Probe Reply F 2 2. Updates Local Data Structures 3. Performs Admission Control Test 4. Computes Allocated Rate of F 2 5. Initiates the RBR Update Phase 6. Allows F 2 to enter the network 11/3/2020 Rowan University F 2 Probe C 1 C 2 C 3 CN Edge 2 Edge 6 F 1 Edge 3 Edge 4 Core Router C 1 discovers that link C 1 -C 2 is congested 1. C 1 initiates the Notification Phase. 2. Edge 1 and Edge 2 adjust allocated rates of F 1 and F 2 to eliminate congestion. Vasil Y. Hnatyshin 44

The BDS Architecture Per-Flow Qo. S Per-Flow Admission Control Excess BW Management Resource Allocation

The BDS Architecture Per-Flow Qo. S Per-Flow Admission Control Excess BW Management Resource Allocation RBR Distribution and Feedback Protocol The BDS Framework Congestion Control Network Architecture Scalability Flow Requirements Definitions of Fairness Specifications and Definitions The BDS Objectives 11/3/2020 The BDS Architecture Rowan University Vasil Y. Hnatyshin 45