Customer Proof of Concept Labs

Transcription

1 Customer Proof of Concept Labs 10 GE and OC-192 Migration with OSPF Auto-Cost Reference and OSPF NSF/SSO CPOC Lab Customer Case Study Plan - Cisco Systems October 2006

2 Legal Notice THE SPECIFICATIONS AND INFORMATION REGARDING THE PRODUCTS IN THIS DOCUMENT ARE SUBJECT TO CHANGE WITHOUT NOTICE. ALL STATEMENTS, INFORMATION, AND RECOMMENDATIONS IN THIS MANUAL ARE BELIEVED TO BE ACCURATE BUT ARE PRESENTED WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED. USERS MUST TAKE FULL RESPONSIBILITY FOR THEIR APPLICATION OF ANY PRODUCTS. THE SOFTWARE LICENSE AND LIMITED WARRANTY FOR THE ACCOMPANYING PRODUCT ARE SET FORTH IN THE INFORMATION PACKET THAT SHIPPED WITH THE PRODUCT AND ARE INCORPORATED HEREIN BY THIS REFERENCE. IF YOU ARE UNABLE TO LOCATE THE SOFTWARE LICENSE OR LIMITED WARRANTY, CONTACT YOUR CISCO REPRESENTATIVE FOR A COPY. The Cisco implementation of TCP header compression is an adaptation of a program developed by the University of California, Berkeley (UCB) as part of UCB s public domain version of the UNIX operating system. All rights reserved. Copyright 1981, Regents of the University of California. NOTWITHSTANDING ANY OTHER WARRANTY HEREIN, ALL DOCUMENT FILES AND SOFTWARE OF THESE SUPPLIERS ARE PROVIDED AS IS WITH ALL FAULTS. CISCO AND THE ABOVE-NAMED SUPPLIERS DISCLAIM ALL WARRANTIES, EXPRESSED OR IMPLIED, INCLUDING, WITHOUT LIMITATION, THOSE OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OR ARISING FROM A COURSE OF DEALING, USAGE, OR TRADE PRACTICE. IN NO EVENT SHALL CISCO OR ITS SUPPLIERS BE LIABLE FOR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, OR INCIDENTAL DAMAGES, INCLUDING, WITHOUT LIMITATION, LOST PROFITS OR LOSS OR DAMAGE TO DATA ARISING OUT OF THE USE OR INABILITY TO USE THIS MANUAL, EVEN IF CISCO OR ITS SUPPLIERS HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. Access Registrar, AccessPath, Any to Any, AtmDirector, Browse with Me, CCDA, CCDE, CCDP, CCIE, CCNA, CCNP, CCSI, CD-PAC, the Cisco logo, Cisco Certified Internetwork Expert logo, CiscoLink, the Cisco Management Connection logo, the Cisco NetWorks logo, the Cisco Powered Network logo, Cisco Systems Capital, the Cisco Systems Capital logo, Cisco Systems Networking Academy, the Cisco Systems Networking Academy logo, the Cisco Technologies logo, ConnectWay, Fast Step, FireRunner, Follow Me Browsing, FormShare, GigaStack, IGX, Intelligence in the Optical Core, Internet Quotient, IP/VC, Kernel Proxy, MGX, MultiPath Data, MultiPath Voice, Natural Network Viewer, NetSonar, Network Registrar, the Networkers logo, Packet, PIX, Point and Click Internetworking, Policy Builder, Precept, ScriptShare, Secure Script, ServiceWay, Shop with Me, SlideCast, SMARTnet, SVX, The Cell, TrafficDirector, TransPath, ViewRunner, Virtual Loop Carrier System, Virtual Service Node, Virtual Voice Line, VisionWay, VlanDirector, Voice LAN, WaRP, Wavelength Router, Wavelength Router Protocol, WebViewer, Workgroup Director, and Workgroup Stack are trademarks; Changing the Way We Work, Live, Play, and Learn, Empowering the Internet Generation, The Internet Economy, and The New Internet Economy are service marks; and ASIST, BPX, Catalyst, Cisco, Cisco IOS, the Cisco IOS logo, Cisco Systems, the Cisco Systems logo, the Cisco Systems Cisco Press logo, Enterprise/Solver, EtherChannel, EtherSwitch, FastHub, FastLink, FastPAD, FastSwitch, GeoTel, IOS, IP/TV, IPX, LightStream, LightSwitch, MICA, NetRanger, Post-Routing, Pre-Routing, Registrar, StrataView Plus, Stratm, TeleRouter, and VCO are registered trademarks of Cisco Systems, Inc. or its affiliates in the U.S. and certain other countries. All other trademarks mentioned in this document are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any of its resellers. Copyright 2007 Cisco Systems, Inc. All rights reserved.

3 Table of Contents EXECUTIVE SUMMARY... 3 TEST OBJECTIVES...3 INTRODUCTION... 4 EQUIPMENT... 5 DESIGN AND TOPOLOGY DIAGRAMS OSPF AREA AUTO-COST TOPOLOGY OSPF PRE-DEFINED AUTO-COST TOPOLOGY TESTING AND RESULTS TEST 1: BANDWIDTH CHANGE ONTO THE EDGE ROUTERS...14 TEST 2: BANDWIDTH CHANGE ONTO THE CORE ROUTERS...20 TEST 3: NON-STOP FORWARDING AND STATEFUL SWITCH-OVER (NSF/SSO)...24 CONCLUSION Page 2 of 27

4 Executive Summary This CPOC visit was a verification of an IGP OSPF design proposal for a large Asian telecommunication company. Due to the increase demand for bandwidth the customer had migrated to a 10 Gigabit Ethernet and OC- 192 links into their network. Consequently, to implement these requirement changes, the customer needed to assess their existing IGP design (OSPF) due to the increase bandwidth where routing decisions were mainly driven by the manual link cost. To address this issue in respect to the scalability and manageability performances, the auto-cost reference was a proposed solution. Furthermore, the customer was highly interested in Non-Stop Forwarding (NSF) and Stateful Switch-Over (SSO) in order to further reduce the possibility of traffic loss during a maintenance course of action when the network upgrades/changes were needed to be executed. Test Objectives This case study includes the following test objectives and a brief description of the test. A summary of these tests for this case study are as follows. Refer to individual test case scenario diagrams for each UUT evaluation. Test 1: Bandwidth Change onto the Edge Routers Scenario A: To verify the uplink bandwidth on one of the Cisco towards the Core was changed with various link bandwidths. Two Edge Cisco 12008s fully meshed with Catalyst 4006; two additional 4006s were connected to the other pair via Gigabit Ethernet. Scenario B: To verify the uplink bandwidths of both Cisco12008 towards the Core devices were changed with various link bandwidths. Two Edge Cisco 12008s fully meshed with Catalyst 4006; an additional 4006s were connected to the other pair via Gigabit Ethernet. Scenario C: To verify the uplink bandwidths of both Cisco12008 towards the Core devices were changed with various link bandwidths. Two Edge Cisco 12008s connected to a pair of Catalyst 4006s. Scenario D: To verify working capabilities that two Edge Cisco 12008s connecting to a single Catalyst 6500 switch; both Cisco 12008s were connected to a single Catalyst Scenario E: To verify working capabilities that two Edge Cisco 12008s connecting to a single Catalyst 6500 switch fully meshed; A Gigabit Ethernet link runs between two Cisco 12008s and Catalyst Test 2: Bandwidth Change into the Core Routers Scenario A: To verify that the two Core 12416s facing the Edge devices worked properly to their bandwidth links. Scenario B: To verify that the two Core 12816s facing the Edge devices worked properly to their bandwidth links. Scenario C: To verify that the two Core 12416s facing the Edge Catalyst 4507R worked properly to their bandwidth links. Test 3: Non-Stop Forwarding and Stateful Switch-Over (NSF/SSO) Scenario A: To verify device DOJ_CR1 with various combination of NSF/SSO turned on/off worked properly. Scenario B: To verify device PS_CR1 with various combination of NSF/SSO turned on/off worked properly. Scenario C: To verify device CHW_LG1 with various combination of NSF/SSO turned on/off worked properly. Page 3 of 27

5 Introduction Prior to the CPOC visit, the customer s network had several different types of links throughout its infrastructure. Also the customer had gone through a 10 Gigabit Ethernet and OC-192 upgrades, as well as, in the Core area along with the Distribution/Edge area that was currently pending. Due to these series of completed and partial upgrades throughout their network, the customer was required to use manual link cost to utilize the new bandwidth upgrade while minimizing their traffic loss. This in turn, had resulted in various complicated and difficult challenges with managing the network administration and maintenances that were required on a periodic basis by the customer. The primary concerns for the customer and CPOC s main objective goals were in verifying the following two criterias: 1. The OSPF auto-cost reference set to 10 Gigabit Ethernet would alleviate the burden of manual link cost assignment throughout the network. (By default, the OSPF will use Fast Ethernet as link cost of 1.) 2. For such tests, they will be conducted while traffic is flowing through the devices to simulate projected traffic loss during the switch over from manual to auto-cost. Page 4 of 27

6 Equipment Equipment and software revisions are detailed in the following table. Cisco Devices Devices Type/Description Engine Type Quantity DUT/Evaluation Chassis wit Fabric 1 PRP (512MB) 2 NSF Test GSR #1 (DOJ_CR1) 12.0(28)S2 2-port OC192- POS(SR) 1-port OC192- POS(SR) 4port OC48- POS(SR) E6 1 To SL_LG2, DJ_CR1 E4+ 1 To NGN_CR1 E4+ 1 To DOJ_GI1 4-port GE E3 1 To DOJ_ID1 Chassis wit Fabric 1 GSR GSR #2 (DOJ_CR2) 12.0(28)S2 PRP (512MB) 1 2-port OC192- POS(SR) 1-port OC192- POS(SR) E6 1 To DJ_CR2, SL_LG1 E4+ 1 To NGN_CR2 4-port OC48- POS(SR) E4+ 1 To DOJ_GI1 4-port GE E3 1 To DOJ_ID1 Chassis wit Fabric 1 GSR #3 (NGN_CR1) 12.0(28)S5 PRP (512MB) 1 2-port OC192- POS(SR) E6 1 To SL_LG1, DOJ_CR1 1-port 10GE E4+ 1 To DJ_CR2 Page 5 of 27

7 Chassis wit Fabric 1 GSR #4 (NGN_CR2) 12.0(28)S5 PRP (512MB) 1 2-port OC192- POS(SR) E6 1 To SL_LG2, DOJ_CR2 1-port 10GE E4+ 1 To DJ_CR1 Chassis wit Fabric 1 GSR #1 (DOJ_GI1) 12.0(26)S6 PRP (512MB) 2 NSF Test 4-port GE E3 1 To Tester 1-port OC48- POS (SR) Chassis wit Fabric E2 2 To DOJ_CR1, DOJ_CR2 1 GRP-B(512MB) 2 NSF Test GSR #2 (PS_CR1) 12.0(25)S3 1-port OC192- POS(SR) E4+ 1 To DOJ_CR1 1-port 10GE(SR) E4+ 1 To NGN_CR2 4-port OC48- POS (SR) E4+ 1 To CHW_LG1 4-port OC3-POS- ISE (SR) E3 1 To CHW_LG2 4-port GE E3 1 To PS_DR1, WS_NAM_LG1, Tester Chassis 1 GRP-B(512MB) 2 NSF Test GSR #3 (PS_CR2) 12.0(25)S3 1-port OC192- POS(SR) E4+ 1 To DOJ_CR2 1-port 10GE(SR) E4+ 1 To NGN_CR1 3-port GE E2 1 To PS_DR1, WS_NAM_LG1,Tester GSR #4 1-port OC48- POS (SR) 4-port OC3-POS (SR) Chassis wit Fabric E2 1 To CHW_LG1 E0 1 To CHW_LG2 1 Page 6 of 27

8 (DJ_CR1) 12.0(25)S3 GRP-B (512MB) 1 1-port OC192- POS (SR) E4+ 2 To DOJ_CR2, NGN_CR1 3-port GE E3 1 To WS_NAM_LG2, CHOONGJU_LG1, Tester Chassis wit Fabric 1 GSR #5 (DJ_CR2) 12.0(25)S3 GRP-B (256MB) 1 1-port OC192- POS (SR) 3-port GE E3 1 E4+ 2 To DOJ_CR1, NGN_CR2 To WS_NAM_LG2, CHOONGJU_LG1, Tester Chassis wit Fabric 1 GSR #1 (CHW_LG2) 12.0(25)S3 GRP-B (256MB) 1 4-Port GE E2 1 To PS_CHW_4006_1, KN_MASAN_4510_1, PS_CHW_FSB1, Tester 1-Port GE E1 1 To CHW_LG1 4-port OC3-POS (SR) Chassis wit Fabric E0 1 1 To PS_CR1, PS_CR2, HFC_MOKPO GRP-B (256MB) 2 NSF Test GSR #2 (CHW_LG1) 12.0(25)S3 1-Port OC48- POS (SR) 4-port GE E2 1 E2 2 To PS_CR1, PS_CR2 To PS_CHW_4006_1, KN_MASAN_4510_1, PS_CHW_FSB1, Tester 1-Port GE E1 1 To CHW_LG1 4-port OC3-POS (SR) E0 1 To HFC_MOKPO C7500 Router C7507 #1 Chassis 1 (DOJ_ID1) d(J) RSP4 or RSP4+ 1 Page 7 of 27

9 GEIP or GEIP+ 2 To DOJ_CR1, DOJ_CR2 FEIP 2 To Tester Chassis 1 C7507#2 (HFC_CHOONGJU) d(J) RSP4 or RSP4+ 1 GEIP or GEIP+ 1 To CHOONGJU_LG1 FEIP 2 To Tester Chassis 1 RSP4 or RSP4+ 1 C7507#3 (PS_DR1) d(J) GEIP or GEIP+ 2 To PS_CR1, PS_CR2 FEIP 2 To Tester VIP PA-OS3-POS (SR) 2 To HFC_MILYANG C7200 Router C7204VXR #1 (HFC_MILYANG) IOS:12.0(19)S4 C7204VXR #2 (HFC_MOKPO) IOS:12.0(19)S4 Chassis 1 NPE-G1 1 To Tester PA-OC3-POS (SR) Chassis 1 1 To PS_DR1 NPE-G1 1 To Tester PA-OC3-POS 2 To CHW_LG1, CHW_LG2 Switch Catalyst 4006#1 (PS_CHW_4006_1) IOS:12.2(25)EWA2 SUP-III 1 48-port GE 1 48-Port FE 2 Catalyst 4006#2 (PS_CHW_4006_2) IOS:12.2(25)EWA2 SUP-III 48-Port GE 1 1 Catalyst 4006#3 SUP-III 1 (KN_MASAN_4006_1) IOS:12.2(25)EWA2 6-Port GE 1 Page 8 of 27

10 Catalyst 4006#4 SUP-III 1 (KN_MASAN_4006_2) IOS:12.2(25)EWA2 48-Port GE 1 Catalyst 4507R#1 SUP-IV 1 (WS_NAM_LG1) IOS:12.2(25)EWA2 6-Port GE 1 Catalyst 4507R#2 SUP-IV 1 (WS_NAM_LG2) IOS:12.2(25)EWA2 48-Port GE 1 Catalyst 4507R#3 SUP-IV 1 (CHOONGJU_LG1) IOS:12.2(25)EWA2 6-Port GE 1 Catalyst 6500 Sup720-3BXL 1 (PS_CHW_FSB1) IOS:12.2(18)SXD5 24-port GE 1 To CHW_LG1, CHW_LG2 L2 Switch 4GE and 12 FE port 3 Ex, Cat3750, Cat3550 Third Party Devices / Software Quantity Device Interfaces Used Software Revision 2 Agilent Traffic Generator GE: 14 Ports FE: 6 Ports 6.5 Build 4.10B Page 9 of 27

11 Design and Topology Diagrams The following topologies were used during various testing. Topology #1 Core Migration Premigration System Figure 1: Physical Network Topology This network diagram represents a portion of customer s future deployed network. It was the intended to include a major portion of Core network and some of the branch deployments as reference. The Cisco GSRs (12816 and 12416) were mainly deployed in the Core portion of the diagram, while Cisco GSR 12008, 7200 and 7500 were deployed in Edge/Distribution area. The Catalyst 4006, 4507R and 6509 represents the Edge portion of the network. Below is additional information that details to the specifications of the network topology: Devices: PE2 and PE3 were identified as PE routers however they were actually functioning as P routers. There were no VRF terminated on them. CR routers represent Core routers. The LG routers represent Local Gateway routers which were linked to the Cores from individual branches. Links: Various links were deployed in the testbed network to provide maximum flexibility and exposure to various links that were actually deployed in customer network. These link types were ranging from Fast Ethernet to 10 Gigabit Ethernet and OC-192. Test Tool and Route Simulation: The Core network OSPF routes: 10,000 routes were injected from Agilent 3. The Internet BGP routes: A total of 240,000 routes were injected from Agilent 1, 3, 4 and 14 with a constant of 30% flapping per minute on various Agilent ports. Page 10 of 27

12 Test Tool and Traffic Flow Generation by Agilent: For OSPF: 2K (LSA 5), 3K (LSA 3) DJ_CR1 (Agilent Port: 3) For BGP: 240K Entries Community Entry Prefix Router Agilent 9318 : : : : : 140 1K 1K 1K 1K 1K 100.x.x.x 110.x.x.x 120.x.x.x 130.x.x.x 140.x.x.x PS_CR : : : : : 190 1K 1K 1K 1K 1K 160.x.x.x 170.x.x.x 180.x.x.x 185.x.x.x 190.x.x.x DJ_CR : 200 8K 200.x.x.x PS_CR : K 210.x.x.x DOJ_Gl : K 220.x.x.x PS_CR2 5 Basic Functionality Baseline and Connectivity Assumptions The basic baseline network connectivity and functionality were successfully accomplished and had been established. Refer to the enclosed Appendix Configuration zip file section for a complete listing of all configurations that were used throughout this test. Page 11 of 27

13 OSPF Area Auto-Cost Topology The following OSPF Auto-Cost testbed topology was used for this test. Topology #2 OSPF Area Auto-Cost System Figure 2: OSPF Area Auto-Cost Topology This illustration represents a portion of customer s network where the Core and selected branches are depicted in shaded color areas. The blue shaded area represents the Core backbone (Area 0), where as, the other color shaded areas depict different geographical regions of customer s branches. Page 12 of 27

14 OSPF Pre-Defined Auto-Cost Topology The following OSPF Pre-Defined Auto-Cost testbed topology was used for this test. Topology #3 Pre-Defined OSPF Cost Auto Cost System Figure 2: OSPF Pre-Defined Auto-Cost Topology Figure 2, represents the customer s current deployed network where the numbers in the black boxes denote a predefined manual link-costs that were required to accurately represent the customer s network specification. The goal of this visit was to simplify their network by substituting the manual link-costs with auto-cost in order to alleviate cost and workload for the network administration personnel. Most notably, the manual link-cost varied considerably throughout their network and consequently this had rendered their network planning/administrating with challenges from a management perspective. Page 13 of 27

15 Testing and Results Procedures and results are given in the following sections. Test 1: Bandwidth Change onto the Edge Routers Theses initial testing scenarios (A E) were to verify the bandwidth on the links of the Edge routers towards the Core, which were modified to represent the various link types: Test 1-Scenario A: To verify the uplink bandwidth on one of the Cisco towards the Core was changed with various link bandwidths. Two Edge Cisco 12008s fully meshed with Catalyst 4006; two additional 4006s were connected to the other pair via Gigabit Ethernet. Test 1-Scenario B: To verify the uplink bandwidths of both Cisco12008 towards the Core devices were changed with various link bandwidths. Two Edge Cisco 12008s fully meshed with Catalyst 4006; an additional 4006s were connected to the other pair via Gigabit Ethernet. Test 1-Scenario C: To verify the uplink bandwidths of both Cisco12008 towards the Core devices were changed with various link bandwidths. Two Edge Cisco 12008s connected to a pair of Catalyst 4006s. Test 1-Scenario D: To verify working capabilities that two Edge Cisco 12008s connecting to a single Catalyst 6500 switch; both Cisco 12008s were connected to a single Catalyst Test 1-Scenario E: To verify working capabilities that two Edge Cisco 12008s connecting to a single Catalyst 6500 switch fully meshed; A Gigabit Ethernet link runs between two Cisco 12008s and Catalyst Procedures The following collective procedures were used for test scenarios A E: 1. Traffic was generated on the Agilent test ports. 2. Continuous ping were generated from neighboring device through the devices for these tests. 3. Auto-cost reference configuration was applied to selected device by auto-cost reference-bandwidth n under OSPF process. 4. The CPU and memory usage was recorded using show cpu/ show proc cpu and show mem/show proc mem CLI commands. 5. Once convergence was verified, traffic was stopped. 6. Traffic loss was measured (actual sent packets received packets) 7. Ping losses were recorded. 8. Random group of prefixes were looked up on the routing table to verify whether the traffic path was changed. 9. (Optional) Depending upon whether traffic loss was observed, manual cost was then re-applied to further reduce any traffic loss. Results and Noted Observations The following data results/observations were recorded, as noted above from the procedures used: Note: For a complete listing of all configurations that were used, refer to the Appendix Configuration to the enclosed zip file. Page 14 of 27

16 Test 1-Scenario A: Two Edge Cisco 12008s fully meshed with Catalyst Two additional 4006s were connected to the other pair via Gigabit Ethernet. Uplink bandwidth on one of Cisco toward core was changed to various link bandwidths. Note: In preparation for this test the following were implemented prior to its testing: Review of Cost application to cope with loop due to BGP default information on the PS Switch. This test was conducted after changing various uplinks to OC-3, OC-48 and GE on CHW_LG2. Figure 3: Test Scenario A Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 3 with the following observations and lessons learned commentaries below: With the OC-3 uplink, cost of link between CHW_LG and 4006 needed to be changed from 50 to 64 or more (70 was used for this test). With the OC-48 uplink, auto-cost 10 of link between CHW_LG1 and CHW_LG2 needed to be changed manually to 5. Cost on GE link with 4006 due to auto-cost will no longer be necessary. With the GE uplink, manual cost needed to be maintained. In case of network consisting of OC-3 links, links toward the 4006 must be manually changed to or above 64 before applying auto-cost. When applying auto-cost, each router needs to be applied sequentially rather than an all-at-once approach. Once auto-cost had been applied, we verified that cost changed was indeed correct, we then move on to the next router It was proven during testing that applying auto-cost sequentially rather than an all-at-once approach had minimized traffic loss. Detailed Test 1-Scenario A Results for Verification: This table provides the criteria overview for Test Scenario A and its final detail results: Page 15 of 27

17 CHW_LG Uplink Traffic Path Changed? CPU (%) Ping Loss (%) Traffic Loss (%) Cost Changed? OC3 POS No GSR: 15~20% 4006: 80~100% loss, 0.11 sec Change Needed GE No GSR : 15~25% 4006: 80~100% loss, sec No Changed Needed OC48 POS Yes Traffic between CHW_LG1 and CHW_LG2 had changed from GE to an OC-48 Path GSR : 15~25% 4006: 80~100% loss, 0.2 sec Change Needed Test 1-Scenario B: Two Edge Cisco 12008s fully meshed with Catalyst An additional 4006s were connected to the other pair via Gigabit Ethernet. Uplink bandwidths of both Cisco12008 toward Core devices were changed to various link bandwidths. Note: In preparation for this test the following were implemented prior to its testing: Review for effectiveness of manual to cope with the loop created by BGP default information on the PS Switch. This test was conducted after changing the uplink bandwidth to OC-48 between CHW_LG1 and CHW_LG2. Figure 4: Test Scenario B Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 4 with the following observations and lessons learned commentaries below: Due to auto-cost, the GE link towards the 4006 did not require manual cost change. Detailed Test 1-Scenario B Results for Verification: Page 16 of 27

18 This table provides the criteria overview for Test Scenario B and its final detail results: CHW_LG Uplink OC48 POS Traffic Path Changed? No CPU (%) Ping Loss (%) Traffic Loss (%) GSR : 15~30% 4006: 65~100% Cost Changed? loss, 0.14 sec No Test 1-Scenario C: Two Edge Cisco 12008s connected to a pair of Catalyst 4006s. Uplink bandwidths of both Cisco12008 toward Core devices were changed to various link bandwidths. Note: In preparation for this test the following were implemented prior to its testing: Review for effectiveness of manual to cope with the loop created by BGP default information on the PS Switch. This test was conducted after changing the uplink bandwidth to OC-48 between CHW_LG1 and CHW_LG2. Figure 5: Test Scenario C Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 5 with the following observations and lessons learned commentaries below: Due to auto-cost, the GE link towards the 4006 did not require manual cost change. When applying auto-cost, it s suggested to start on the Edge (CAT 4006) to prevent changes in traffic path. Therefore, auto-cost should be applied on the CAT 4006, then to the CHW_LG. Detailed Test 1-Scenario C Results for Verification: This table provides the criteria overview for Test Scenario C and its final detail results: Page 17 of 27

19 CHW_LG Uplink OC48 POS Traffic Path Changed? No CPU (%) Ping Loss (%) Traffic Loss (%) GSR : 15~30% 4006: 65~100% Cost Changed? 0 8 loss, 0% No Test 1-Scenario D: Two Edge Cisco 12008s connecting to a single Catalyst 6500 switch. Both Cisco 12008s were connected to a single Catalyst Note: In preparation for this test the following were implemented prior to its testing: Link bandwidth between CHW_LG and PS_CR1 was set to OC-48; traffic then would be sent to PS_CHW_FBS. Link cost between CHW_LG1a and PS_CR1 will then be changed to OC-48. Auto-Cost applied to CHW_LG only (not onto PS_CR) Link cost between CHW_LG and PS_CR set manually to 10. GE link between CHW_LG and PS_CHW_FBS set manually to 3. GE links between CHW_LGs set manually to 3. GE links between CHW_LG and 4006 no longer require manual cost (CHW and WS however, should keep manual cost of 50. Figure 6: Test Scenario D Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 6 with the following observations and lessons learned commentaries below: After Auto-Cost had been applied: Page 18 of 27

20 Manual cost to 10 between CHW_LG and PS_CR were deleted (auto-cost of 4 kept). GE between CHW_LG and PS_CHW_FBS keeps manual cost of three. GE Links in between CHW_LG keeps manual cost of three. Test 1-Scenario E: Two Edge Cisco 12008s connecting to a single Catalyst 6500 switch fully meshed. Gigabit Ethernet link runs between two Cisco 12008s and Catalyst Note: In preparation for this test the following were implemented prior to its testing: Link bandwidth between CHW_LG and PS_CR1 was set to OC-48; traffic then would be sent to PS_CHW_FBS. Link cost between CHW_LG1a and PS_CR1 will then be changed to OC-48. Auto-Cost applied to CHW_LG only (not onto PS_CR) Link cost between CHW_LG and PS_CR set manually to 10. GE link between CHW_LG and PS_CHW_FBS set manually to 3. GE links between CHW_LGs set manually to 3. GE links between CHW_LG and 4006 no longer require manual cost (CHW and WS however, should keep manual cost of 50. Figure 7: Test Scenario E Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 7 with the following observations and lessons learned commentaries below: After Auto-Cost had been applied: Manual cost to 10 between CHW_LG and PS_CR were deleted (auto-cost of 4 kept). GE between CHW_LG and PS_CHW_FBS keeps manual cost of three. GE Links in between CHW_LG keeps manual cost of three. Page 19 of 27

21 Test 2: Bandwidth Change onto the Core Routers Theses initial testing scenarios (A C) were to verify the bandwidth on the links of the Core routers towards the Edge devices, which were modified to represent the various link types: Test 2-Scenario A: To verify that the two Core 12416s facing the Edge devices worked properly to their bandwidth links. Test 2-Scenario B: To verify that the two Core 12816s facing the Edge devices worked properly to their bandwidth links. Test 2-Scenario C: To verify that the two Core 12416s facing the Edge Catalyst 4507R worked properly to their bandwidth links. Procedures The following collective procedures were used for test scenarios A C: 1. Traffic was generated on the Agilent test ports. 2. Continuous ping were generated from neighboring device through the devices for these tests. 3. Auto-cost reference configuration was applied to selected device by auto-cost reference-bandwidth n under OSPF process. 4. The CPU and memory usage was recorded using show cpu/ show proc cpu and show mem/show proc mem CLI commands. 5. Once convergence was verified, traffic was stopped. 6. Traffic loss was measured (actual sent packets received packets) 7. Ping losses were recorded. 8. Random group of prefixes were looked up on the routing table to verify whether the traffic path was changed. 9. (Optional) Depending upon whether traffic loss was observed, manual cost was then re-applied to further reduce any traffic loss. Results and Noted Observations The following data results/observations were recorded, as noted above from the procedures used: Note: For a complete listing of all configurations that were used, refer to the Appendix Configuration to the enclosed zip file. Test 2-Scenario A: Two Core 12416s facing the Edge devices worked properly to their bandwidth links. Note: In preparation for this test the following were implemented prior to its testing: Review of manual cost assigning to prevent traffic re-routing through WS_NAM_LG1. Manual cost applied after changing the link between PS_CR and WS_NAM to OC-48. Page 20 of 27

22 Figure 8: Test Scenario A Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 8 with the following observations and lessons learned commentaries below: Even after auto-cost had been applied, the link between the PS_CR and S_NAM had required a manual cost of 50. Also after PS_CR had auto-cost applied, the traffic between PS_CR had taken the path of the PS_DR instead of the OC-192 in the backbone area (lower cost). Therefore, the traffic path had not changed. Detailed Test 2-Scenario A Results for Verification: This table provides the criteria overview for Test Scenario A and its final detail results: PS_CR WS_NAM Link OC48 POS Traffic Path Changed? No CPU (%) Ping Loss (%) Traffic Loss (%) ~15% Cost Changed? 0 0 No Page 21 of 27

23 Test 2-Scenario B: Two Core 12816s facing the Edge devices worked properly to their bandwidth links. Note: In preparation for this test the following were implemented prior to its testing: Reconfigured the auto-cost between DOJ_GI1, DOJ_CR1 and DOJ_CR2. Verified that there was no traffic path change after applying auto-cost and deleting manual cost between DOJ_CR1 and DOJ_CR2. Figure 9: Test Scenario B Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 9 with the following observations and lessons learned commentaries below: Manual cost or 10 was no longer necessary after auto-cost was applied. There was no path or CPU change after deleting manual cost of 10. Also, there was no path or CPU change on DOJ_CR after auto-cost was applied. Detailed Test 2-Scenario B Results for Verification: This table provides the criteria overview for Test Scenario B and its final detail results: CHW_LG Uplink Traffic Path Changed? CPU (%) Ping Loss (%) Traffic Loss (%) Cost Changed? OC48 POS No 0% 0 0 No Page 22 of 27

24 Test 2-Scenario C: Two Core 12416s facing the Edge Catalyst 4507R worked properly to their bandwidth links. Note: In preparation for this test the following were implemented prior to its testing: Verified the effect of auto-cost on all CR and LG routers (including route-path and CPU utilization). Figure 10: Test Scenario C Diagram In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 10 with the following observations and lessons learned commentaries below: When auto-cost was applied sequentially, traffic path changed was observed also small traffic loss was observed. CPU utilization had fluctuated approximately between 30-50% due to the amount of LSAs on PS_CHW (approximately with a fluctuation of %). Detailed Test 2-Scenario C Results for Verification: This table provides the criteria overview for Test Scenario C and its final detail results: Traffic Path Changed? Path Change was observed due to autocost application on 2 GE links sequentially CPU (%) Ping Loss (%) Traffic Loss (%) Cost Changed? 30-50% Loss, 0.1 Sec No Page 23 of 27

25 Test 3: Non-Stop Forwarding and Stateful Switch-Over (NSF/SSO) Theses initial testing scenarios (A C) were to verify various combinations of NSF/SSO while being turned on or off onto specified Core and Edge devices. Test Environment Test 3-Scenario A: To verify DOJ_CR1 with various combinations of NSF/SSO turned on/off and record its results/observations. Test 3-Scenario B: To verify PS_CR1 with various combinations of NSF/SSO turn on/off and record its results/observations. Test 3-Scenario C: To verify CHW_LG1 with various combinations of NSF/SSO turn on/off and record its results/observations. Topology #1 Core Migration Premigration System was used for all three (A-C) testing scenarios. The UUT were CHW_LG1, PS_CR1 and DOJ_CR1 during these NSF/SSO enable and disable feature testing as depicted in figure 11 below by the yellow shaded area circles onto the respective devices/links. The NSF configuration application order was: DOJ_CR1 PS_CR1 CHW_LG1. The traffic flow was: DJ_CR2 PS_CHW_FBS1. Note: In preparation for this test the following were implemented prior to its testing: All routers were configured for NSF/SSO mode in redundancy configuration and verified the BGP/OSPF NSF operation status through the CLI command show ip bgp neighbor, show ip ospf and debug commands. The route processor was performed on some of the routers (DJ_CR1, DJ_CR2, DOJ_GI1 and DOJ_CR1). Figure11: Test Scenario A-C NSF/SSO Diagram Page 24 of 27

26 In this testing scenario, the UUT were all the devices and links within the yellow shaded area, depicted in figure 11 with the following observations and lessons learned commentaries below: The following is a list of CLI commands that were used for examining the CPU, memory and NSF operation after the route processor switchover through the CLI command redundancy forced-switchover and RP OIR. show CPU, show proc CPU show memory, show proc mem show redundancy show redundancy states show redundancy clients show redundancy history debug redundancy Refer to the Appendix Configuration of the enclosed zip file for a complete CLI command output listing. Procedures The following collective procedures were used for test scenarios A C: 1. Verified the current redundancy configuration, memory and IOS version in all routers RPs (Active and Standby RP). 2. Agilent and SMB generated background unicast traffic to go through the routers. 3. Issued the show commands listed above and recorded the logged the output information to a file. 4. All routers were set as NSF mode in BGP and OSPF by the CLI command bgp gracefulrestart and NSF. 5. For BGP NSF, BGP neighbor was reset to establish new BGP session with NSF capability. 6. Performed the show command show ip bgp neighbor to verify NSF ran well. 7. Issued the show commands listed above and logged the output information to a file. 8. Reset the standby RP CLI command redundancy forced-switchover. 9. Verified the RP and Line Card online-up time and any error logs, and then recorded them. 10. Monitored traffic from Agilent and SMB, along with ping loss rate and record them. 11. Issued the CLI show commands listed above and logged the output information to a file compared to Step 3 results. 12. Reset the standby RP by RP OIR and also performed Steps 9, 10 and 11 again. 13. For OSPF NSF, issued the CLI show command show ip ospf to verify NSF had ran properly. 14. Issued the show commands listed above and logged the output information to a file. 15. Reset the standby RP by issuing CLI command redundancy forced-switchover. 16. Verified the RP and Line Card online-up time and any error logs, and then recorded them. 17. Monitored the traffic from Agilent and SMB and ping loss rate and then recorded them. 18. Issued the CLI show commands listed above and logged the output information to a file compared to Step 3 results. 19. Reset the standby RP by RP OIR and also performed Steps 16, 17 and 18 again. Results and Noted Observations The following data results/observations were recorded, as noted above from the procedures used: Page 25 of 27

27 Note: For a complete listing of all configurations that were used, refer to the Appendix Configuration to the enclosed zip file. Detailed Test 3-Scenarios A C Results for Verification: This table provides the criteria overview for Test Scenarios A C and its final detail results: GSR List (UUT) DOJ_CR1 PS_CR1 CHW_LG1 IOS/RP Test Method Traffic Loss Noted Observations 12.0(28)S2 PRP 12.0(28)S3 GRP 12.0(25)S3 GRP OSPF OFF/BGPOFF 50 Sec Loss Due to OSPF Converging Faster than BGP, Traffic Loss would Increase. OSPF ON/BGP OFF 25 Sec Loss Completion of BGP Update becomes Shorter due to OSPF NSF feature. OSPF ON/BGP ON 0 Sec Loss Optimum Result; No Losses OSPF OFF/BGP OFF OSPF ON/BGP OFF 48 Sec Loss 22 Sec Loss Due to OSPF Default Information from DOJ_CR, a Loop was Detected on PS_CR. However, traffic from PS_CR to DOJ_CR had no losses due to Default Information. OSPF ON/BGP ON 0 Sec Loss Optimum Result; No Losses OSPF OFF/BGP OFF OSPF ON/BGP OFF 23 Sec Loss 17 Sec Loss Traffic toward DJ_CR2 suffered minimal loss when OSPF NSF was turned on and default information was originated. OSPF ON/BGP ON 0 Sec Loss Optimum Result; No Losses Page 26 of 27

28 Conclusion By in large the proposed CPOC solution for the customer s network in addressing their different types of links throughout its infrastructure was satisfactory and positive. Most notably was in resolving the customer s auto-cost concerns in dealing with their 10 Gigabit Ethernet and OC-192 upgrades, as well as, in the Core area along with the Distribution/Edge area that was currently pending. Also by overcoming some of their completed and partial current network upgrades throughout their network system; we were able to demonstrate to the customer the use of auto-cost reference and NSF/SSO for managing and meeting their traffic requirements. This had resolved the customer s frequent challenges in managing their network administration and maintenances that were required and on-going by the customer. CPOC s main objective goals were met and successfully accomplished in assessing the customer s primary concerns in verifying the following criterias in: The OSPF auto-cost reference set to 10 Gigabit Ethernet alleviated the burden of manual link cost assignment throughout the network. These tests were performed and accomplished while traffic was flowing through the devices in order to simulate projected traffic loss during the switch over from manual to auto-cost per customer requirements and validation. Page 27 of 27