Your SlideShare is downloading. ×
Executive Summary.doc.doc
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Executive Summary.doc.doc

463

Published on

Published in: Business, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
463
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
7
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. QoS Design, Deployment & Monitoring for Networks Page 1 of 17
  • 2. QoS Design, Deployment & Monitoring for Networks Table Of Contents: Executive Summary.......................................................................................4 QoS for Networks..........................................................................................4 What is QoS...............................................................................................4 Why QoS Management................................................................................5 Metrics and Levels of Scaling........................................................................5 NJ Edge........................................................................................................6 MPLS...........................................................................................................6 QoS with NJEdge for Video.............................................................................6 Factors which Degrade Video Conferencing Quality.........................................6 Factors which Degrade Streaming Video Quality.............................................7 QoS with NJEdge for Voice..............................................................................7 Factors which Degrade Voice Quality.............................................................7 Packet loss..............................................................................................7 Factors which Degrade Voice Quality.............................................................7 Class Of Service Layer 2...........................................................................8 Type Of Service Layer 3............................................................................8 Video Classification for NJEdge.....................................................................9 Video......................................................................................................9 Video conferencing...................................................................................9 Streaming video (IP/TV)...........................................................................9 3rd party video partners...........................................................................9 Voice Classification for NJEdge......................................................................9 Voice......................................................................................................9 VoIP control channels...............................................................................9 VoIP RTP bearer channels........................................................................10 QoS-QPM Scalability.....................................................................................10 IP Telephony Templates.............................................................................10 Change Control & Verification.....................................................................10 Deployment Target Options........................................................................10 Cross-Database Update.............................................................................10 Test Environment and Assumptions................................................................11 Network...................................................................................................11 QoS-QPM Deployment Scope of Work ............................................................12 QoS Design...........................................................................................12 QoS Deployment....................................................................................13 QoS User Acceptance Testing...................................................................14 QoS -QPM Training & Documentation........................................................14 Scenarios.................................................................................................14 Scalable QoS Monitoring Technologies............................................................15 Pricing and Scope........................................................................................15 Small-Medium Sized Network.....................................................................15 Definition..............................................................................................15 Scope ..................................................................................................15 Estimated Hours required for Completion .................................................16 Large Network..........................................................................................16 Definition..............................................................................................16 Scope...................................................................................................16 Estimated Hours required for Completion .................................................16 Page 2 of 17
  • 3. QoS Design, Deployment & Monitoring for Networks Very Large Network...................................................................................17 Definition..............................................................................................17 Scope...................................................................................................17 Estimated Hours required for Completion .................................................17 Page 3 of 17
  • 4. QoS Design, Deployment & Monitoring for Networks QPM - QoS Design & Deployment Template Executive Summary As applications such as voice and video appear on converging networks, the need for Quality of Service (QoS) from nearly every device in the network becomes crucial. As a result, in large networks the provisioning and administration of QoS quickly become daunting and complex tasks. Cisco Quality of Service Policy Manager (QPM) is designed to provision QoS for voice, video, and data networks and has many useful features to facilitate scaling. A main feature is the set of IP Telephony templates, which embed the Cisco QoS recommendations for voice into predefined policy device-groups. This simplifies an administrator's task to simply assigning the devices and interfaces to the corresponding group and clicking on the Deploy button. Furthermore, these templates can be tailored to customer requirements. Such modifications are included in the test scenarios detailed in this paper. Additionally, QPM offers features such as rollback (to quickly undo changes that have been pushed out to network devices) and also device verification (which allows an administrator to determine quickly whether the deployed QoS configurations have been tampered with). There are four levels of scaling are defined: Small-scale IP Telephony deployments—fewer than 500 IP Phones or 2000 switch ports Medium-scale IP Telephony deployments—500 to 2500 IP Phones or 2000 to 5000 switch ports Large-scale IP Telephony deployments—2500 to 10,000 IP Phones or 5000 to 15,000 switch ports Very-large-scale IP Telephony deployments—more than 10,000 IP Phones or more than 15,000 switch ports QoS for Networks What is QoS Customer networks exist to service application requirements and end users efficiently. The tremendous growth of the Internet and corporate intranets, the wide variety of new bandwidth-hungry applications, and convergence of data, voice, and video traffic over consolidated IP infrastructures has had a major impact on the ability of networks to provide predictable, measurable, and guaranteed services to these applications. Achieving the required Quality of Service (QoS) through the proper management of network delays, bandwidth requirements, and packet loss parameters, while maintaining simplicity, scalability, and manageability of the network is the fundamental solution to running an infrastructure that serves business applications end-to-end. Cisco IOS Software offers a portfolio of Page 4 of 17
  • 5. QoS Design, Deployment & Monitoring for Networks QoS features that enable customer networks to address voice, video, and data application requirements, and are extensively deployed by numerous enterprises and Service Provider networks today. Why QoS Management A critical enabler for the convergence of voice, video, and data onto a single network is QoS. QoS technologies ensure that latency- and jitter-intolerant applications, such as voice and video, will receive end-to-end priority service. Achieving end-to-end priority services requires QoS functionality from virtually every device in the network. QoS is required from nearly every device in the network. This requirement, coupled with the size and variation found in typical enterprise networks, establishes a strong case for centralized QoS management. Cisco QPM is the tool of choice for centralizing, provisioning, and deploying QoS to both LAN and WAN network devices for voice, video, and data. Metrics and Levels of Scaling How well does QPM scale? QoS deployments incorporate many different factors, including: Number of devices provisioned Number of interfaces and ports per device Number of policies per interface Number of commands required per policy Simply using the number of devices as a reference is misleading because provisioning a 240-port Cisco Catalyst® 6000 device will require more commands than a Cisco 1750 Router with two interfaces. Totaling the number of interfaces is also insufficient because more complex QoS policies might apply to certain critical interfaces within the network. Additionally, the raw number of interfaces and ports within an enterprise's network is not a well-known number and thus wouldn't prove meaningful. This metric could be further broken down into the number of IP Phones deployed in the central campus (LAN) and the number of IP Phones deployed at remote sites (primarily WAN). Examining typical customer deployments reveals that the approximate ratio of campus IP Phones to remote-site IP Phones varies between 65 and 80 percent. Using the number of IP Phones deployed as a metric, QoS for IP Telephony deployments can be categorized into four distinct levels of scale: Small-scale deployments—fewer than 500 IP Phones Medium-scale deployments—500 to 2500 IP Phones Large-scale deployments—2500 to 10,000 IP Phones Very-large-scale deployments—more than 10,000 IP Phones Page 5 of 17
  • 6. QoS Design, Deployment & Monitoring for Networks NJ Edge NJ Edge is a broadband statewide network designed to enhance the teaching, research and public service missions of New Jersey's colleges and universities. The range of capabilities, resources and services offer economies of scale, provide expanded opportunities for integrating emerging technologies, and promote new forms of inter-institutional collaboration. This private, statewide infrastructure effectively "raises the bar" for high performance data and video capabilities across the state's three LATA boundaries and extends the reach of higher education to off- campus learners, K-12, as well as corporate and community constituents. Data and video equipment vendors include Cisco, Polycom and First Virtual Communications. The custom IP-VPN network for NJEDge.Net deploys Multiprotocol Label Switching (MPLS) technology across the state. MPLS Multi-Protocol Label Switching (MPLS) defines a mechanism for packet forwarding in network routers. It was originally developed to provide faster packet forwarding than traditional IP routing, although improvements in router hardware have reduced the importance of speed in packet forwarding. However, the flexibility of MPLS has led to it becoming the default way for modern networks to achieve Quality of Service (QoS), next generation VPN services, and optical signaling. Traditional IP networks are connectionless: when a packet is received, the router determines the next hop using the destination IP address on the packet alongside information from its own forwarding table. The router's forwarding tables contain information on the network topology. They use an IP routing protocol, such as OSPF, IS-IS, BGP, RIP or static configuration, to keep their information synchronized with changes in the network. MPLS also uses IP addresses, either v4 or v6, to identify end points and intermediate switches and routers. This makes MPLS networks IP-compatible and easily integrated with traditional IP networks. However, unlike traditional IP, MPLS flows are connection-oriented and packets are routed along pre-configured Label Switched Paths (LSPs). QoS with NJEdge for Video Factors which Degrade Video Conferencing Quality Unlike voice, video has a very high, extremely variable packet rate Much higher average MTU Queuing :The LLQ will fill to capacity regularly ; Queuing delay = serialization delay as utilization approaches 100% End-to-end delay 200 msec target delay budget Page 6 of 17
  • 7. QoS Design, Deployment & Monitoring for Networks Jitter buffer 20-70 msec Factors which Degrade Streaming Video Quality Has a very high, extremely variable packet rate Much higher average MTU Queuing:Because of the tolerance for e-2-e delay, streaming video should go into a bw-based queue End-to-end delay 4–5 secs Jitter buffer 1 MB (read long latency tolerance) QoS with NJEdge for Voice Factors which Degrade Voice Quality Packet loss Current Cisco GW DSP CODEC algorithms can correct for 30 msec of lost voice—1 G.729A voice packet contains 20 msec of voice.Lost packets induce “clipping” and temporarily expand the jitter buffer, which increases end-to-end latency. One lost FAX over IP packet causes a MODEM retrain;2 drops cause a call disconnect Causes of packet loss: Network quality, network congestion and delay variation (jitter bufferunder-runs) Factors which Degrade Voice Quality Cisco GW DSPs Uses an Adaptive Jitter Buffer which Only Has 10 msec of “Extra” Buffer. Packet Dropped If Instantaneous Jitter Is > 10 msec End-to-End Delay: ITU G.114 states one-way delay <= 150 msec ~200 msec is acceptable CODEC -G.729A = 25 msec (20msec+5msec look ahead) Queuing-Queuing delay = serialization delay as utilization approaches 100% Serialization Propagation and network delay-6.3 usec/km + network delay (variable) Jitter buffer 20-50 msec Page 7 of 17
  • 8. QoS Design, Deployment & Monitoring for Networks Class Of Service Layer 2 Type Of Service Layer 3 Page 8 of 17
  • 9. QoS Design, Deployment & Monitoring for Networks Video Classification for NJEdge Video Video conferencing CoS = 4, IP Prec = 4, DSCP = AF41 Streaming video (IP/TV) CoS = 1, IP Prec = 1, DSCP = AF13—Recommended for enterprises UDP = IP/TV 3.2 provides customer port configuration 3rd party video partners VCON—Can set ToS PictureTel—Can set ToS Polycom—Can set ToS RadVision—Can not set ToS Voice Classification for NJEdge Voice VoIP control channels CoS = 3, IP Prec = 3, DSCP = AF31 Page 9 of 17
  • 10. QoS Design, Deployment & Monitoring for Networks VoIP RTP bearer channels CoS = 5, IP Prec = 5, DSCP = EF QoS-QPM Scalability IP Telephony Templates To simplify and expedite the deployment of the recommended QoS features, QPM has predefined policy templates that are an integral part of QPM. These templates contain all the IP Telephony QoS recommendations for classification, trust, queueing, shaping, and fragmentation and correspond to the type and role of the device or interface in the network. Provisioning complex QoS commands throughout the network quickly becomes a challenging, complex, and time-consuming task. However, with QPM, all an administrator has to do is add or import devices to QPM, put the right interfaces into the corresponding device groups in the IP Telephony template, save the database, and then deploy the policy database. Change Control & Verification QPM offers important mechanisms to facilitate large-scale QoS rollout. Device rollback allows an administrator to quickly revert to any previous QPM deployment state. This is similar in function to the Undo button in a word processor. Device verification is an important tool because sometimes QoS configurations are changed without the administrator's consent or knowledge. Verification is an option that QPM presents to allow for a quick check of the configurations to ensure that they are all intact as originally deployed. If the configurations are not the same, then a mismatch is reported, and the correct commands can be deployed quickly. Deployment Target Options QPM offers the options of sending the deployment commands directly to the device or to a log file (for Trivial File Transfer Protocol [TFTP] or other deployment tool distribution), or to both. These options make it easy to integrate QPM into enterprises that already have their own deployment mechanisms tailored to their specific needs. Cross-Database Update As this paper will show, some scales of deployment are administered better by using multiple QPM databases. When multiple databases are used, a change in one QPM database is confined to that database only. To prevent having to make the identical changes to every database being administered, QPM provides a cross-database update function. When a user needs to modify a QPM database, he has the option of updating this change on some (or all) of the other databases being administered. Page 10 of 17
  • 11. QoS Design, Deployment & Monitoring for Networks For instance, in most scenarios, initially only QoS for IP Telephony is rolled out. Then the enterprise decides to expand its QoS provisioning to include a mission-critical data application. This mission-critical Enterprise Resource Planning (ERP) application is to be colored at both the server and client ends of the network, in addition to being guaranteed bandwidth over the WAN. To make these changes, an administrator only has to manually update one QPM database and then use the cross-database update function to push these new modifications to all other databases. Test Environment and Assumptions The general network model for IP Telephony is shown in Figure 4 & Figure 5. Network Figure 4: Small To Medium Scale IP Telephony Lab Network Page 11 of 17
  • 12. QoS Design, Deployment & Monitoring for Networks Figure 5: Large-Scale IP Telephony Lab Network QPM testing in labs provided valuable insight into how well QPM scales. However, because some large enterprises are engaged in IP Telephony deployments well beyond even the scale of these networks above, another approach to large-scale QPM testing is required. QoS-QPM Deployment Scope of Work The QoS-QPM deployment for small to medium network can be phased put into three different phases: QoS Design QoS Deployment QoS User Acceptance Testing QoS -QPM Training & Documentation QoS Design Designing involves setting the right policies at the right place. The definitions of Class Map, Policy Map and the Service Policies are defined in this phase. LLQ is used as the queuing mechanism for the Voice and Class Based Weighted Fair Queue (CBWFQ) is used as the queuing mechanism for Data. Definitions of Marking, Mapping, Queuing, Policing are created in this phase tailored according to the network type and size. Page 12 of 17
  • 13. QoS Design, Deployment & Monitoring for Networks QPM (QoS Policy Manager) is used for Enterprise QoS deployment and monitoring. The there most major design aspects while designing the QoS are: Classification—Marking the packet with a specific priority denoting a requirement for special service from the network Scheduling—Assigning packets to one of multiple queues (based on classification) for expedited treatment through the network Provisioning—Accurately calculating the required bandwidth for all applications plus element overhead QoS Deployment Small- and medium-scale deployments should be administered using a single QPM database. RME import is optional if fewer than 10 devices are being administered using QPM. Separate QPM databases are recommended for large-scale IPT deployments: one database for campus (LAN) devices and remote-site (primarily WAN) devices. Central site WAN edge routers should be included in the WAN database (despite their location at the central campus). Very-large-scale scenarios are to be administered by QPM as modular large-scale scenarios. Each large-scale scenario has a separate campus and WAN QPM database. To increase from large-scale to very-large-scale scenarios, separate QPM databases should be created for every 10,000 IP Phones deployed. The ratio of campus IP Phones to WAN IP Phones within the enterprise should be kept consistent. For example, if the number of campus IP Phones is 75 percent of the total, then a separate campus database should be created for every 7500 campus IP Phones, and a separate WAN database should be created for every 2500 IP Phones. These recommendations are summarized in Table 10. Table 10: Recommendations for QoS Scaling Using QPM Scale of Deployment RME QPM Database Strategy Cross- Import Databas e Update Small Scale (Fewer Than 500 IP Optional Single Enterprise No Phones or 2000 switch ports) Database Medium Scale (500 to 2500 IP Mandatory Single Enterprise No Phones or 2000 to 5000 Switch Database Ports) Large Scale (2500 to 10,000 IP Mandatory Single Campus DB + No Phones or 5000 to 15000 Switch Single WAN DB ports) Page 13 of 17
  • 14. QoS Design, Deployment & Monitoring for Networks Very Large-Scale (More Than Mandatory 1 Campus Database Yes 10,000 IP Phones or more than per 7500 15000 switch ports) Campus IP Phones or 10000 switch ports 1 WAN Database per Yes 2500 Remote IP Phones/5000 switch ports QoS User Acceptance Testing The User Acceptance Testing (UAT) will be done at the end of deployment of the QoS enterprise wise. Sniffer traces will be collected at every critical nodes or points and the traces will be analyzed to check the QoS settings are working as per design with the proper marking for specific applications as determined in the design phase. QoS -QPM Training & Documentation Training will be provided to client identified network administrators for operational and day to day administrative aspects of the QPM(Cisco QoS Policy Manager) Server. At the end of the engagement we will also provide a detailed documentation with troubleshooting guidelines. Scenarios The scenarios were as follows: each scale of enterprise (small, medium, large, very large) would be provisioned for IP Telephony QoS using QPM on each of the servers used in these tests. These scenarios would include not only deploying the QoS but also rolling back the deployment and performing a worst-case scenario device verification. Then, the enterprise is re provisioned to support QoS not only for IP Telephony but also to protect a mission-critical ERP application. (In this case the application is SAP, identified by the well-known TCP port 3200.) The modifications include these changes: SAP traffic is marked to Differentiated Services Code Point (DCSP) 28 (AF32) by the servers in the campus data center to mark server-to-client traffic. SAP traffic is marked to DSCP 28 (AF32) by every single Catalyst 6000 port connected to an IP Phone or PC to mark client-to-server traffic. SAP traffic is marked to DSCP 28 (AF32) by the Fast Ethernet router interfaces nearest the clients for segments with switches that do not support DSCP marking to mark client-to-server traffic. Page 14 of 17
  • 15. QoS Design, Deployment & Monitoring for Networks SAP traffic is provisioned for CBWFQ with a minimum bandwidth guarantee of 25 percent on all WAN links throughout the enterprise. These changes are made to the QPM database (and for very-large-scale scenarios cross-database updates are made). When all database operations have been completed, the new policies are deployed to the entire enterprise. Scalable QoS Monitoring Technologies QoS management does not end with QoS policy configuration. Monitoring the effectiveness of the policies is crucial. Monitoring must be conducted immediately after QoS policy deployment (to ensure completeness and functionality) and routinely afterward (to ensure that changing network conditions do not disrupt QoS levels). Cisco scalable QoS monitoring technologies include: Class-based QoS Management Information Base (MIB) Service Assurance Agent NetFlow switching and collecting Pricing and Scope For pricing and scope we have categorized the QoS deployment into 3 categories. They are: Small-Medium Sized Network Large Network Very Large Network Small-Medium Sized Network Definition Small-Medium Sized Network typically will have less than 2500 IP phones or 5000 switch ports and with less than 3 links to the NJ Edge or any other remote sites. Scope 1. Install the OS(Windows 2000 Server/Advanced Server/Professional). 2. Apply the necessary security patches to the OS. 3. Install the Anti Virus Software on the server. 4. Install & Configure the Cisco Works Common Services on the server. 5. Install QPM (Cisco QoS Policy Manager) on the server. 6. Run Device discovery to discover all the network devices. 7. Define QoS Policy like Policy for Voice, Video, Data etc. 8. Define QoS Deployment Group. 9. Define QoS Policy Group for LAN and WAN. Page 15 of 17
  • 16. QoS Design, Deployment & Monitoring for Networks 10. Deploy the QoS Policy on the Access Layer. 11. Deploy the QoS Policy on the Collapsed Core Layer. 12. Deploy the QoS Policy on the WAN to NJ Edge. 13. Sniff and capture packets at different layers with a network sniffer to get the colors of the packets as they traverse the network. 14. Complete the UAT(User Acceptance Testing). 15. Training & Documentation. Estimated Hours required for Completion The total number of hours required to complete will be estimated about 40 hours- 50 hours depending upon the complexity of the LAN and WAN. Large Network Definition Large Sized Network typically will have less than 10,000 IP phones or 15,000 switch ports and with less than 6 links to the NJ Edge or any other remote sites. Scope 1. Install the OS(Windows 2000 Server/Advanced Server/Professional). 2. Apply the necessary security patches to the OS. 3. Install the Anti Virus Software on the server. 4. Install & Configure the Cisco Works Common Services on the server. 5. Install QPM (Cisco QoS Policy Manager) on the server. 6. Run Device discovery to discover all the network devices. 7. Define QoS Policy like Policy for Voice, Video, Data etc. 8. Define QoS Deployment Group. 9. Define QoS Policy Group for LAN and WAN. 10. Run “NBAR” application discovery protocol to learn about different application running on the network for example ERP, Citrix, Napstar, Kaaza etc. 11. Define the coloring mechanism with values for different type of traffic identified while defining QoS Policy. 12. Deploy the QoS Policy on the Access Layer. 13. Deploy the QoS Policy on the Distribution Layer. 14. Deploy the QoS Policy on the Core Layer. 15. Deploy the QoS Policy on the WAN to NJ Edge. 16. Sniff and capture packets at different layers with a network sniffer to get the colors of the packets as they traverse the network. 17. Complete the UAT(User Acceptance Testing). 18. Training & Documentation. Estimated Hours required for Completion The total number of hours required to complete will be estimated about 70 hours- 80 hours depending upon the complexity of the LAN and WAN. Page 16 of 17
  • 17. QoS Design, Deployment & Monitoring for Networks Very Large Network Definition Very Large Sized Network typically will have greater than 10,000 IP phones or 15,000 switch ports and with greater than 6 and less than 10 links to the NJ Edge or any other remote sites. Scope 1. Install the OS(Windows 2000 Server/Advanced Server/Professional). 2. Apply the necessary security patches to the OS. 3. Install the Anti Virus Software on the server. 4. Install & Configure the Cisco Works Common Services on the server. 5. Install QPM (Cisco QoS Policy Manager) on the server. 6. Run Device discovery to discover all the network devices. 7. Define QoS Policy like Policy for Voice, Video, Data etc. 8. Define QoS Deployment Group. 9. Define QoS Policy Group for LAN and WAN. 10. Run “NBAR” application discovery protocol to learn about different application running on the network for example ERP, Citrix, Napstar, Kaaza etc. 11. Define the coloring mechanism with values for different type of traffic identified while defining QoS Policy. 12. Deploy the QoS Policy on the Access Layer. 13. Deploy the QoS Policy on the Distribution Layer. 14. Deploy the QoS Policy on the Core Layer. 15. Deploy the QoS Policy on the WAN to NJ Edge. 16. Sniff and capture packets at different layers with a network sniffer to get the colors of the packets as they traverse the network. 17. Complete the UAT(User Acceptance Testing). 18. Training & Documentation. Estimated Hours required for Completion The total number of hours required to complete will be estimated about 90 hours- 100 hours depending upon the complexity of the LAN and WAN and also the number of WAN or remote links to the NJ Edge and other remote sites. Page 17 of 17

×