Summary Report Alcatel-Lucent Triple Play Service Delivery Architecture (TPSDA) Validation Report

Summary Report Alcatel-Lucent Triple Play Service Delivery Architecture (TPSDA) Validation Report Residential Subscriber and Enterprise Service Scaling Non-Stop Service Verification QoS for Triple Play Services Quality Measurement for IPTV Service Delivery Isocore Internetworking Lab Isocore Technical Document Reference: ITD-12084 Version (v1.1): 02/15/07 Isocore Internetworking Lab 1

12359 Sunrise Valley Drive, STE 100 Reston, VA 20191 Phone: 703.860.1777 Fax: 703.860.1682 www.isocore.com 2007 by ISOCORE. All rights reserved. For more information about the testing capabilities of the Internetworking Lab, please email testing@isocore.com This document should not be reproduced as a certification or a validation reference from ISOCORE. The reproduction of this document to the third party requires written approval from ISOCORE Corporation. Isocore Internetworking Lab 2

EXECUTIVE SUMMARY This summary report presents the results of a comprehensive triple play service delivery architecture (TPSDA) validation and performance verification conducted independently by Isocore. The test series focused on an extensive evaluation of Alcatel-Lucent s feature set related to large-scale residential triple play deployments with enhanced subscriber management while supporting a very high number of mixed enterprise virtual private network (VPN) services over a single Internet Protocol, MultiProtocol Label Switching (IP/MPLS) network. Testing was conducted with the Alcatel-Lucent 7750 Service Router (SR) and the 7450 Ethernet Service Switch (ESS). Both products performed exceptionally well under extreme scaling conditions imposed by the test series. Alcatel-Lucent demonstrated a flexible end-to-end solution for a variety of deployment methods necessary in increasing Quality of Experience of any subscriber in a large scale triple play environment. KEY FINDINGS The Alcatel-Lucent TPSDA, 7750 SR and 7450 ESS platforms and related triple play feature set exceeded the triple play requirements set forth by Isocore for this test series. The key findings of the evaluation include: The base network scaled to 120,000 active subscriber hosts and more than 43,000 enterprise class Layer 2 and Layer 3 VPN services. The enhanced subscriber management (ESM) feature set performed the process of subscriber identification and instantiation of policies and resources across all test scenarios. Traffic forwarding to all residential and enterprise services were not impacted beyond the microsecond level when control processor modules (CPMs) were failed over. The security feature set demonstrated the network was secure and no intrusion attempts were successful. The QoS feature set of Alcatel-Lucent 7750 SR and 7450 ESS platforms is able to maintain subscriber Service Level Agreements (SLA) under severe congestion and separate subscribers at the forwarding plane level. The systems were stable under considerable stress through numerous iteration and test cycles. Based on this comprehensive evaluation it is clear to Isocore that Alcatel-Lucent offers a complete endto-end triple play solution. The 7750 SR and 7450 ESS that form the core of the solution support a scalable mix of residential triple play and enterprise services. The consistency of results produced through multiple iterations of testing assures the non-stop delivery of all services. The end result of testing validated that what was presented in the lab is deployable. Isocore Internetworking Lab 3

1 TEST OVERVIEW AND FOCUS As part of its advanced triple play and IPTV certification tests Isocore evaluated Alcatel-Lucent s triple play service delivery solution for its ability to meet the strict requirements of a scaled residential triple play deployment. Isocore s evaluation focused on Alcatel-Lucent s TPSDA and the ability of the 7450 ESS and 7750 SR to support scaled residential services. Although residential triple play services were the primary focus, a highly-scaled deployment of Layer 2 and Layer 3 VPN services was added to the topology to demonstrate that the network and platforms are capable of carrying both residential triple play services and enterprise class VPN services simultaneously. The base network included 120,000 active subscriber hosts and more than 43,000 enterprise class Layer 2 and Layer 3 VPN services. Table 1 provides a summary of the significant scale achieved during the event. Under these highly-scaled conditions, Isocore performed a thorough analysis of all aspects of residential triple play service delivery. This included an in-depth evaluation of Alcatel-Lucent s subscriber management, hierarchical quality of service (H-QoS), security, subscriber host authentication, highavailability, resiliency feature, and operations administration and maintenance (OAM), as well as a thorough analysis of a flexible architecture. Table1: Summary of the Test Results Capability Validation Residential Service Scaling - Number of Subscribers/ Hosts verified on the Triple Play Setup Enterprise Service Scaling With Residential Services Performance Benchmarking Highlights of the Triple Play Setup Description Network totals o 80,000 active subscribers o 120,000 active subscriber hosts o 80,000 active SLA policies o 120,000 anti-spoofing filters o 120,000 ingress and egress queues o 120,000 ingress and egress schedulers o 80k 802.1Q VLAN tags Network Totals 43,500 total Layer 2 and Layer 3 enterprise VPN services including: o 2,500 BGP/ MPLS VPNs o 4,000 virtual private LAN service (VPLS) instances o 37,000 Point-to-Point Ethernet virtual leased lines (VLLs) o Including 2,000 Ethernet VLL backhauls terminated into BGP/MPLS VPN Services Non-stop service verification: Minimal traffic loss to residential subscribers for duration of 9.92µs 11.52 µs with high-availability Traffic prioritization of VoIP, VoD and broadcast TV (BTV) service. No traffic loss for VoIP, VOD, and BTV traffic under heavy congestion No packet loss during long duration traffic forwarding tests (10-12 hrs run) This test was the first of its kind in the industry to focus on validating the ability to deliver a highly-scaled mix of residential triple play services and enterprise VPN services. The challenge of a test like this is the multi-faceted and complex nature of the effort. The execution included careful design and preparation and hundreds of hours of actual testing. The level of stress imposed on the network, architecture, and systems under test was significant. Isocore Internetworking Lab 4

Alcatel-Lucent TPSDA Performance Verification Testing - Executive Summary The end result validated that what was presented in the lab is truly deployable. Numerous iteration and test cycles were performed and the systems were extremely stable under considerable stress. The related triple play ESM feature set is comprehensive, assuring that residential triple play service deployments are highly manageable. The systems and architecture can support stringent SLAs per subscriber and provide the security required for residential services. The high-availability and related resiliency feature set assured the reliability of the residential and VPN services. In short, the AlcatelLucent TPSDA, 7750 SR and 7450 ESS platforms and related triple play feature set exceeded the triple play requirements set forth by Isocore for this test series. 1.1 TEST METHODOLOGY AND SYSTEM TEST SETUP Figure 1 provides an overview of the test network setup. The setup used two, fully loaded 7450 ESS-7s as broadband service aggregators (BSAs). Each of the 7450 ESS-7s was equipped with nine 10-port Gigabit Ethernet media dependent adapters (MDAs) and one 2-port 10 Gigabit Ethernet MDA. BSAs aggregated traffic from directly connected DSLAMs using a VPLS-based TPSDA. Two 7750 SR-12s were setup as broadband service routers (BSRs) to aggregate the BSAs and support direct connections from DSLAMs when an aggregation network is not required. For this event the BSR used a routed TPSDA model for DSLAM aggregation. The BSRs were equipped with a variety of 10/100/1000 Gigabit and 10 Gigabit Ethernet MDAs. A 7750 SR-7 was used as a gateway for content delivery. The Agilent N2X Multiservice Test Solution was used as the source for highspeed Internet (HSI), BTV multicast content, vdeo on demand (VOD), voice over IP (VoIP), and signaling all active subscriber dynamic host configuration protocol (DHCP) sessions through the DHCP server. All 7450 ESS and 7750 SRs were running SR_OS version 4.0R6. The Agilent N2x also emulated all hosts on the network. Emulated DSLAM and Subscribers DSL Modem IP DSLAM... 25,000 subscribers 45,000 hosts 15,000 subscribers using RHG (single MAC/IP DHCP) Routed Aggregation Model DSL Modem Agilent N2X IP DSLAM 4 GbE Links... Emulated DSLAM and Subscribers VPLS Infrastructure for Residential TP Services 22 GbE Links Emulated DSLAM and Subscribers IP DSLAM BSR-2 TP Gateway Router 22 GbE Links BSA-12... Routed Aggregation Model Agilent N2X 25,000 subscribers 45,000 hosts Agilent N2X BSR-1 Business Services BSR 4K VPLS, 2.5K L3VPNs, 28K VLLs BSAs 4K VPLS, 11K VLLs VPLS Aggregation Model Agilent N2X DSL Modem Emulated BTV, VoIP, HSI,VoD Traffic Source BSA-11 DSL Modem 4 GbE Links IP DSLAM 10GbE Links 1 GbE Links BSA (11,12) - 7450 ESS-7 BSR (1,2) - 7750 SR-12 TP Gateway Router - 7750 SR-7... 15,000 subscribers using RHG (single MAC/IP DHCP) Emulated DSLAM and Subscribers Figure 1: Alcatel-Lucent TPSDA Verification Test Bed Isocore selected the Agilent N2X platform to verify the scalability of the Alcatel-Lucent TPSDA solution because it met the stringent test requirements put forth by Isocore. In order to execute a test of this magnitude and collect and analyze the significant amount of statistics generated, automated test execution was a requirement. Automation was used for all aspects of the test, including highly scaled DHCP client setup, subscriber emulations, DSLAM emulation, mixed service traffic generation, test Isocore Internetworking Lab 5

measurements, statistics collection and analysis as well as media delivery index (MDI) measurements across tens of thousands of BTV receivers. To further facilitate test execution, a customized GUI (test manager) was created. 2 TESTING DETAILS AND RESULTS 2.1 RESIDENTIAL TRIPLE PLAY SERVICE DELIVERY SCALING The objective of the first phase of testing was to perform a detailed analysis of a broad range of requirements to support scaled residential triple play deployments. These tests included the validation of scalability, flexibility and robustness of the Alcatel-Lucent TPSDA architecture, triple play ESM and other related triple play features and components of the 7750 SR and 7450 ESS. This scaled test setup was used throughout the event for staging, test evaluation, and all test cases that followed, such as HA, resiliency, QoS, security, and OAM. To show flexibility of the architecture and feature set, the test was designed to verify Alcatel-Lucent s TPSDA ability to support a variety of deployment models. Alcatel-Lucent refers to the flexibility to support various deployment options as Any Mode of Operation (AMO). The test included the following Alcatel- Lucent TPSDA AMO models: VPLS-based aggregation model Routed aggregation model VLAN-per-service model each service is on a separate VLAN and each triple play subscriber s traffic is spread over three separate VLANs VLAN-per-subscriber model each subscriber is on a separate VLAN and each triple play subscriber s traffic from all three services is carried by a single VLAN Routed home gateway (RHG) - CPE presents single IP/MAC pair and DHCP session per subscriber Bridged home gateway (BHG) - CPE presents multiple IP/MAC pair and DHCP session per subscriber Alcatel-Lucent s ESM feature set automates subscriber identification, classification and policy instantiation for residential service deployments. ESM is used to identify residential subscribers by parsing the various DHCP option fields of the authenticated DHCP ACK message offered by the DHCP server. Once a residential subscriber is identified, it is assigned a subscriber ID, the associated subscriber SLA and security policies are applied and resources such as queues and schedulers are allocated. Regardless of whether a VPLS-based or routed architecture is used, the subscriber management functionality must be transparent. It is clear to Isocore that Alcatel-Lucent has achieved this requirement. The Agilent N2X signaled and maintained all active DHCP residential subscriber hosts in the network. In both VPLS and routed architectures, the BSA or BSR performed DHCP snooping on the initial DHCP request and relayed the request to the DHCP server. The Agilent N2X also emulated the DSLAM behavior by applying the Option 82 circuit ID, remote ID and Option 60 vendor equipment identifier to all DHCP requests. When the outbound authenticated DHCP ACK was received on the BSA or BSR, it was again snooped and the Option 82 fields were parsed to identify the subscriber type and create a unique subscriber ID per subscriber. Isocore Internetworking Lab 6

To emulate real world residential subscribers, three types of subscriber profiles were used. All subscribers were represented by the VLAN per subscriber model and every subscriber was managed by Alcatel- Lucent s ESM. ESM dynamically provisioned each subscriber with a subscriber-id, subscriber policies, SLA policies, anti-spoofing filters and allocated the required resources. The three profiles that were used included: o Subscriber A Receiving only HSI and BTV traffic o Subscriber B Receiving only HSI, BTV and VoIP o Subscriber C Receiving HSI, BTV, VoIP, VoD services The Agilent N2X was used to source all BTV multicast content. The network carried 400 standard definition (SD) channels and 100 high definition (HD) channels. For the VPLS-based aggregation model the BTV content was delivered over a separate VPLS instance as a VLAN per service model. Alcatel- Lucent s VPLS implementation supports IGMP snooping, so only joined multicast groups are forwarded to the end DSLAM emulated by the Agilent N2X. The routed model received BTV content from the routed PIM domain on the BSRs. Both of these models adhere to the DSL Forum s TR-101 and TR-102 standards for multicast delivery. On the subscriber access side of the network, the VLAN-per-service model was used to deliver BTV to the Agilent N2X emulated DSLAMs. Each of the 52 Gigabit Ethernet access ports joined 125 IGMP HD and SD channels, representing a total of 6,500 channels of multicast content delivered to the DSLAMs. Every subscriber on the network received high-speed Internet (HSI) traffic and every type B and type C subscriber received simulated VoIP traffic. Additionally, 9% of type C subscribers received simulated video on demand (VoD) traffic. Throughout the test event, traffic flows were used to validate subscriber state, network stability and network availability. Traffic flows were run repeatedly overnight and demonstrated low latency with no packet loss throughout the event. Traffic forwarding verification was performed using the advanced statistics collection ability of the Agilent N2X. MDI measurements were used to measure the quality of experience (QoE) of the BTV multicast content received by the DSLAM. MDI measurements included MDI delay factor and MDI media loss rate. Media delivery index is typically represented by two numbers delay factor (DF) and media loss rate (MLR) DF:MLR. Measured MDI value for the setup was 4.90:0 DF:MLR, indicating low jitter with no packet loss. Additionally, latency measurements, and packet loss measurements of unicast traffic flows were made throughout the test event. During staging and test execution all 120,000 DHCP sessions were established and released many times. During each cycle all hosts were authenticated to the network, SLA and security policies were applied, and traffic flows were generated to all subscribers. During removal of sessions, all DHCP states were released and policies were removed and resources were de-allocated. Many instances of starting and stopping multicast traffic with thousands of IGMP joins and leaves hitting the network were executed during all phases of testing. These activities imposed significant stress on the system under test, but at no time did the systems exhibit any signs of instability or performance degradation. In order to validate flexibility of the ESM feature set DHCP option 82 circuit-id, remote-id as well as option 67 (applied by the DHCP server) were parsed and used to uniquely identify subscribers. Alcatel- Lucent s 7750 SR and 7450 ESS also support the ability to mark DHCP option 82 fields. To add additional stress, the BSAs and BSRs were configured to overwrite all inbound option 82 fields received from the Agilent N2X. Isocore spent considerable time evaluating the process of subscriber identification and instantiation of policies and resources. The ESM feature set performed these functions consistently during hundreds of iterations and across all test scenarios. Throughout the test event the network architecture, platforms and ESM feature set proved to be very stable under considerable scale and stress. Due to the complexity of the event and time constraints, testing to maximum scale of any particular area was beyond the scope of this event. It was evident by the systems performance and available resources this test could have scaled higher. Isocore Internetworking Lab 7

Table 2 summarizes the performance and scaling results of the triple play residential services scaling test. Table 2: Results of the Residential TPSDA Scaling Test Triple Play Residential Service Delivery Scaling Network Totals 80,000 active subscribers and policies o 50,000 active subscribers and policies for VPLS-based TPSDA o 30,000 active subscribers and policies for routed aggregation model 120,000 active subscriber hosts, anti-spoofing filters, Ingress/ Egress Queues and Schedulers o 90,000 active DHCP hosts, filters, queues and schedulers for VPLS-based TPSDA o 30,000 active DHCP hosts, filters, queues and schedulers for routed aggregation model 80,000 802.1Q VLAN tags o 50,000 802.1Q VLAN tags for VPLS-based TPSDA o 30,000 802.1Q VLAN tags for routed aggregation model VPLS-Based TPSDA Aggregation Model (Per BSA statistics) 25,000 residential subscribers per BSA o 10,000 RHG with single MAC/IP pair and DHCP session per BSA o 15,000 BHG per BSA with 2 to 3 MAC/IP pair and DHCP session 45,000 active DHCP hosts per BSA 45,000 ingress and egress queues per BSA 25,000 ingress and egress schedulers per BSA 25,000 VLAN tags per BSA BTV multicast content delivery: o Separate VPLS instance used as a VLAN per service model for BTV delivery o 400 channels of SD and 100 channels of HD multicast content on VPLS service o 22 GE ports per BSA, 44 total GE ports receiving 125 channel mix of SD and HD content o Total IGMP joins present 5,500 for the VPLS VLAN per service TPSDA Routed Aggregation Model (Per BSR statistics) 15,000 subscribers per BSR using RHG (single MAC/IP DHCP) 15,000 active DHCP hosts per BSR 15,000 ingress and egress queues per BSR 15,000 ingress and egress schedulers per BSR 15,000 VLAN tags per BSR BTV multicast content delivery: o Routed PIM using a VLAN-per-service model for BTV delivery o 400 channels of SD and 100 channels of HD multicast content on VPLS service o 6 GE ports per BSA, 12 total GE ports receiving 125 channel mix of SD and HD content o Total IGMP joins present 1,500 for the Routed VLAN per service TPSDA Results o No traffic loss observed for any of the VoIP, HSI, and VoD services during overnight runs (10 to 12 hour runs) o Systems demonstrated utmost stability o Low MDI values 4.90:0 (DF:MLR) 2.2 ENTERPRISE SERVICES SCALING (SIMULTANEOUS LAYER 2 AND LAYER 3 SERVICES) The objective of this test was to verify the capability of Alcatel-Lucent 7750 SR, and 7450 ESS to concurrently support large scale enterprise VPN and residential triple play services. This test was built upon the baseline test validated in section 2.1. The test involved building upon the residential service configuration a mix of Layer 3 and Layer 2 VPN services across the 7750 SRs and 7450 ESSs. Isocore Internetworking Lab 8

Table 3 summarizes the results of the scaled enterprise services test. This demonstrates a very high degree of scaling, especially in combination with the large scale residential triple play deployment. All configured enterprise services on both the 7750 SR, and 7450 ESS were operational with traffic for all iterations of the residential triple play test scenarios including multiple overnight runs without traffic loss. Table 3: Results of the Enterprise Service Scaling Tests Layer 3/ Layer 2 Enterprise Service Scalability 34,500 Enterprise Services across BSRs o 4,000 VPLS Services o 2,500 L3 VPN Services o 28,000 point-to-point Ethernet VLLs 15,000 Enterprise Services across BSAs o 4,000 VPLS Service o 9,000 point-to-point Ethernet VLLs o 2,000 backhaul Ethernet VLLs terminated at BSRs Layer 3 VPN Service No Traffic loss observed to any services 2.3 HIGH AVAILABILITY FOR RESIDENTIAL AND ENTERPRISE SERVICES The Alcatel-Lucent 7750 SR and 7450 ESS support an extensive High Availability (HA) feature set which includes non-stop routing for protocols and non-stop services. In the event of CPM failure the 7450 ESS and 7750 SR do not lose protocol adjacencies; network routes and all services remain operational. For residential services this includes maintaining the DHCP states of residential subscriber hosts, policy and resource instantiation, and multicast membership states. The test involved failing over an active CPM and verifying the state of all active network protocols, the residential subscribers, and enterprise class VPN services. Both the 7450 ESS and 7750 SR CPMs were failed repeatedly during the test. The test was performed with all 120,000 active subscriber hosts and more than 43,000 VPN services operational and all related traffic flows running. The test confirmed traffic forwarding to all residential and enterprise services was not impacted beyond the microsecond level when CPMs were failed over. Isocore confirms the non-stop service and non-stop routing capability of the Alcatel-Lucent 7750 SR, and 7450 ESS. Table 4: Results of High-Availability Verification High-Availability Verification No loss of any of the 120,000 active subscriber hosts when either a BSR or BSA CPM was failed-over No loss of any of the 43,000 enterprise services when either a BSR or BSA CPM was failed-over Worst case traffic (HSI, VoIP, and VoD) disruption period, when BSR is failed 11.52µs (refer figure 1) Worst case traffic (HSI, VoIP, and VoD) disruption period, when BSA is failed 9.92µs (refer figure 1) Consistent low MDI DF:MLR value - less than 5.05:0.23 for 400 SD channels 2.3.1 RESIDENTIAL SERVICE ASSURANCE OF TPSDA USING MPLS BASED FAST REROUTE TECHNIQUE This test verified the resiliency of Alcatel-Lucent s TPSDA using MPLS-based fast reroute techniques. The goal of the test was to measure the extent of traffic loss experienced by residential and VPN services when a forced link or node failure scenario was conducted. Multiple tests of link and node failures were conducted. The worst case traffic loss measurement for an MPLS fast reroute link failure test was 16.38ms. The node failure scenario involved failing BSR1 (see figure 1). It was expected that subscribers attached to BSR1 would be lost, but any residential or enterprise class traffic traversing BSR1 would failover in sub 50ms range. The worst case traffic loss observed due to BSR1 (node) failure was 16.12ms. No visible signs of performance degradation were observed on the Alcatel-Lucent 7750 SR or 7450 ESS during the execution of link failure and node failure Isocore Internetworking Lab 9

scenarios under highly scaled conditions. The systems under test showed flawless failure recovery and demonstrated high resiliency of residential and enterprise services well below the industry recommended standards. 2.3.2 PERSISTENCE FOR RESIDENTIAL TRIPLE PLAY SERVICES The Alcatel-Lucent 7750 SR and 7450 ESS supports a persistence feature in which all active subscriber DHCP states are maintained on the system s compact flash memory. In the event of a full system reboot, the system recovers all subscriber state information that was available prior to the reboot. This prevents subscribers from having to re-establish authentication and reacquire DHCP state to the network. This test was performed on a 7450 ESS BSA with 45,000 active subscriber hosts. Test traffic was sent to all active hosts and the node was power cycled. After the boot sequence, all DHCP states were operational and all traffic flows were re-established without any intervention. This feature greatly improves the convergence of residential services and prevents a high number of clients re-authenticating to the network. 2.3.3 TRAFFIC PRIORITIZATION AND QUALITY OF SERVICE (RESIDENTIAL SUBSCRIBER SLA TEST) The objective of this test was to evaluate the hierarchal QoS feature set of Alcatel-Lucent 7750 SR and 7450 ESS platforms. This test validated the ability to maintain subscriber SLAs under severe congestion as well as separation of all subscribers at the forwarding plane level. The residential subscriber SLA test was configured as per Figure 2. All policies and resources were applied by Alcatel-Lucent s ESM. The test included 556 subscribers of type C on a Gigabit Ethernet port. It involved all subscribers receiving VoIP, HSI traffic and 10% receiving VoD traffic, along with the presence of 125 SD and HD multicast BTV traffic on the GE port. Figure 2 represents the subscriber SLA test setup and traffic load. Throughout the event all subscribers had ingress and egress SLA policies similar to Fig 2. SLA Test Setup Egress traffic toward DSLAM GE VLAN-1 (sub-1) VLAN-n (sub-n) VoIP CIR/PIR 100kbps FC=EF VOD CIR/PIR 2.5Mbps FC=H2 HSI CIR 0 PIR max FC=BE VoIP CIR/PIR 100kbps FC=EF VOD CIR/PIR 2.5Mbps FC=H2 HSI CIR 0 PIR max FC=BE Sub-1 Scheduler CIR 2.6Mbps PIR 2.7Mbps Sub-n Scheduler CIR 2.6Mbps PIR 2.7Mbps VLAN per Subscriber Model 556 Subscriber Type C 1,668 Sub hosts presented by BHG 3 egress queues per sub 1 scheduler per sub 1,668 total egress queues 556 Egress schedulers total Test demonstrated no loss and low latency of all priority traffic MDI of BTV 4.88:0 (DF:MLR) Figure 2: SLA Prioritization Test Setup The test represented a Gigabit Ethernet attached DSLAM with a significant number of subscribers and a comprehensive mix of traffic flows. The measurements were made to verify the ability to prioritize the voice, VOD and BTV traffic while the egress Gigabit Ethernet port was heavily over-subscribed with HSI best effort traffic. Figure 3 shows screen captures from the Agilent N2X system showing that Alcatel- Lucent 7450 ESS was able to prioritize the VoIP and VOD traffic over HSI traffic when the egress access Isocore Internetworking Lab 10

link was heavily oversubscribed. Note that the average latency in Figure 3 was measured across the network including the gateway router, BSR and BSA. Also note that the receive test throughput in Figure 3 includes the additional VLAN tag overhead. Figure 3: Results of traffic prioritization, Alcatel-Lucent 7450 ESS dropping best effort traffic for oversubscribed links There was no traffic loss and latency was low for all high priority traffic. Additionally, measured MDI values were also very low in the range of 4.88:0.0 (DF: MLR). This test validated that even under severe congestion SLAs of high priority traffic can be maintained. It also demonstrated the critical requirement for separation on a per subscriber service basis for residential triple play service deployments. 2.4 TPSDA SERVICE SECURITY FEATURE SET Alcatel-Lucent s 7750 SR and 7450 ESS support a comprehensive security feature set for securing residential triple play services. These tests were designed to verify the security feature set available for Alcatel-Lucent s TPSDA. For both routed and VPLS-based aggregation the ESM automatically applies an anti-spoofing filter to secure each subscriber. The anti-spoofing filter is based on the snooping of the authenticated DHCP ACK IP and MAC information. For VPLS-based TPSDA the security feature supports a split-horizon between residential subscribers to prevent residents accessing one another across the Layer 2 bridged domain. The platforms also support a MAC pinning feature which prevents a MAC learned on one VLAN from being spoofed somewhere else in the network. A MAC protect feature is used to secure gateway and other critical MACs learned in a VPLS domain from being spoofed. Multiple tests were run to try and gain unauthorized access to the network, to attempt to spoof MAC learning from various points in the network and to spoof the gateway MAC and create a black hole to the network. Both IP and non-ip traffic streams were constructed to offer false MAC information to the network. In all cases the network was secure and no intrusion attempts were successful. 2.5 TPSDA OAM TOOLS Throughout the test event numerous OAM tools were used to test, troubleshoot and manage the network. Tools such as OAM service ping, MAC ping, and VCCV ping were used to verify and troubleshoot the setup. The ability of ESM to identify individual subscribers and allow the operator to monitor the subscriber hosts state and traffic statistics proved very useful during the tests. During the tests, the service mirroring feature was also extensively used to troubleshoot and validate the behavior on the network. A single mirror destination was setup on the 7750 SR and a protocol analyzer tool was attached. In this way, we were able to mirror traffic from any point in the network back to the centralized mirror destination. Service mirroring was a critical tool during all phases of the event. Isocore Internetworking Lab 11

During the test, TPSDA subscriber host connectivity verification (SHCV) OAM tool was also verified. The SHCV tool uses a periodic ARP to test the host connectivity emulated by the Agilent N2X. SHCV periodically monitors the state of a subscriber host and generates an alarm if the subscriber host is down. 2.6 TPSDA SERVICE PROVISIONING AND SUBSCRIBER MANAGEMENT TOOLS Isocore was presented with a separate demonstration of the Alcatel-Lucent 5620 Service Aware Manager (SAM) and 5750 Subscriber Service Controller (SSC). The 5620 SAM is a management suite that provides element management, end-to-end service provisioning and fault management of the network infrastructure and can be fully integrated into any OSS infrastructure. The 5750 SSC is a tool that manages residential subscribers and services for any size of triple play deployment. The 5620 SAM was used during portions of the overall test event and was also included in the demonstration of 5750 SSC. The demonstration included a basic triple play deployment setup as a case study and presented the 5620 SAM s intuitive GUI that provides end-to-end provisioning and management of residential and enterprise service deployments. Examples of testing and troubleshooting with the OAM tools and fault management applications were shown during the demonstration. An overview of user configurable templates and wizards was also presented. From the capabilities shown during the session, it was evident that the 5620 SAM can greatly simplify the provisioning and management of both residential and enterprise VPN service deployments The Alcatel-Lucent 5750 SSC is a centralized policy control element for managing residential subscribers. The demonstration showed that the 5750 SSC is integrated with the 5620 SAM, billing systems and network elements. The SSC offers capabilities such as subscriber authentication and identification, management of subscriber policies, subscriber web portal, and admission control for VoD services. it also includes an integrated DHCP server for identifying and authenticating the subscriber to the network. The demonstration highlighted how the SSC pushed relevant SLAs and security policies to the network based on the subscription profile created during the demonstration. It further demonstrated how, using a Web portal, subscribers can manage or upgrade their services allowing zero-touch management of residential triple play services. The observations made during this demonstration confirm that with the use of the 5620 SAM and the 5750 SSC service providers can fully utilize the scalability and flexibility of the Alcatel-Lucent TPSDA. The capabilities shown simplify triple play service deployments through the automation of a broad range of critical functions and they are an important part of the Alcatel-Lucent residential triple play portfolio. 3 CONCLUSIONS Based on this comprehensive evaluation it is clear that Alcatel-Lucent offers a complete end-to-end triple play solution. The 7750 SR and 7450 ESS that form the core of the solution are highly scalable, extremely stable and provide a comprehensive triple play feature set that is required for any real world triple play deployment. Both VPLS-based and IP-routed triple play service delivery architectures are equally sound in meeting residential triple play requirements. Alcatel-Lucent s support of any mode of operation for TPSDA provided the flexibility to evaluate different deployment combinations during the test series. The overall results indicate that the tested solution is one of the most comprehensive triple play service delivery solutions available in the industry. The consistency of results produced through multiple iterations of tests conducted, and very large configurations built to support high numbers of residential subscribers assures the non-stop delivery of residential and enterprise services. Throughout the test event, the conditions under which these tests were conducted repeatedly confirmed the reliability of the systems under test. This was evidenced by uptime of the systems. From the time the equipment was racked, through more than a week of staging and hundreds of iterations of test setup and tear down and 12 straight days of continuous testing, all Isocore Internetworking Lab 12

systems remained operational and stable. All nodes exceeded 21 days of system uptime with the exception of the two reboots executed late in the test event to conduct the node failure scenario of a BSR for MPLS FRR and a BSA for the persistence test. In conclusion, Isocore feels extremely comfortable in stating the scalability and feature richness of Alcatel-Lucent TPSDA solution and strongly believes, based on certifiable results, that the setup evaluated during this test series is fully deployable in its current form for delivering comprehensive triple play services in any network environment. 4 LIST OF ABBREVIATIONS AMO Any Mode Of Operation BGP Border Gateway Protocol BHG Bridged Home Gateway BSA Broadband Service Aggregation Router BSR Broadband Service Router BTV Broadcast Television CIR Committed Information Rate CPE Customer Premise Equipment CPM Control Processor Module DF Delay Factor DHCP Dynamic Host Configuration Protocol DLSAM Digital Subscriber Line Access Multiplexer ESM Enhanced Subscriber Management ESS Ethernet Service Switch GE Gigabit Ethernet GUI Graphical User Interface HA High Availability HD High Definition H-QoS Hierarchical Quality Of Service HSI High Speed Internet IGMP Internet Group Management Protocol IP Internet Protocol IPTV IP Television L2 Layer 2 L3 Layer 3 MAC Media Access Control MDI Media Delivery Index MLR Media Loss Rate MPLS Multi-Protocol Label Switching OAM Operation And Management OSS Open Source Solution PIR Peak Information Rate QoS Quality Of Service RHG Routed Home Gateway Isocore Internetworking Lab 13

SAM SD SF SHCV SLA SR SSC TP TPSDA VCCV VLL VOD VoIP VPLS Service Aware Manager Standard Definition Switch Fabric Subscriber Host Connectivity Verification Service Level Agreement Service Router Subscriber Service Controller Triple Play Triple-Play Service Delivery Architecture Virtual Circuit Connectivity Verification Virtual Leased Line Service Video On Demand Voice Over IP Virtual Private LAN Service About Isocore Internetworking Lab: Isocore has a fully equipped, state of the art, Internetworking Laboratory which conducts technology validation and certification, as well as product evaluation. The Lab's current focus is on evaluating triple play deployment solutions including IPTV service deployment architecture validation and design. Additionally, Isocore is the leader in validation and interoperability of emerging and next generation technologies such as Carrier Ethernet, IPv6, IP Optical Integration, wireless backhauling and Layer 2/3 Virtual Private Networks (VPNs). Isocore has already verified several solutions and products next generation network deployments. Major router and switch vendors, Service Providers, and test equipment suppliers participate in lab activities. Isocore Internetworking Lab 14