|Issue:||North America 2011|
|Topic:||Ethernet backhaul migration Is your backhaul network ready for Ethernet?|
|Title:||Chief Technology Officer.|
Vikas Arora is the Chief Technology Officer at EXFO. Mr Arora’s 18 years of technical and business leadership experience spans IP voice, video, Ethernet, optical networks, service assurance and network management. Mr Arora was a co-founder and CTO of iPine Networks, later acquired by Nakina. At Nortel, Mr Arora held senior management and architect positions within the OpenIP and optical business units. Vikas Arora earned a Bachelor of Technology from G.B. Pant University and a Master of Science in Computer Science from the University of Saskatchewan.
The smartphone, the social Internet and multimedia applications – and the traffic they generate – are forcing wireless service operators to adopt packet-based Ethernet/IP in their access and core networks. Ethernet reduces both CAPEX and OPEX and provides economically scalable backhaul to deal with rising bandwidth requirements. Nevertheless, migrating to Ethernet/IP-centric backhaul networks is a challenge, especially with regard to maintaining QoS to maintain quality and reliability for each service. The use of appropriate test strategies helps guarantee network performance.
Is your backhaul network ready for Ethernet? by Vikas Arora, Chief Technology Officer, EXFO Service lifecycle testing for QoS and QoE The growing demand of mobile data-centric services is driven by smartphones, the rollout of 3G/4G next-generation mobile broadband networks, and the popularity of social media (e.g., Facebook, Twitter, etc.) and multimedia applications (e.g., gaming, YouTube, etc.). To meet this demand, operators must upgrade their mobile backhaul networks. With 25 per cent to 30 per cent of their OPEX allocated to network operations – most of which is dedicated to backhaul networks – operators are looking at ways to migrate to more scalable and cost-efficient technologies to support future bandwidth requirements and protect future investments. IP/Ethernet solutions are not only seen as new opportunities, but also as new challenges since operators need to validate, assess and guarantee the performance and the quality of their services. Unlike traditional circuit switched networks, IP/Ethernet networks are efficient at using the available capacity of the network. However, they also introduce new impairments, which affect the QoS and QoE. While applications can manage with some loss, their overall quality will suffer if that amount of loss becomes too high. The same applies to other key performance indicators (KPIs), such as latency and jitter. With this in mind, operators must change the way they test their networks; performance testing is no longer enough. With IP/Ethernet, both network and service performance must be tested. This means that they must be equipped to manage the user experience across the network in a scalable manner, no matter the architecture. As operators move toward packet-based IP/Ethernet backhaul networks, they must: 1) have the right tools, 2) have the right field and operations crews, and 3) have the capacity to align with the IP/Ethernet service activation requirements – all of which across the full network and service lifecycle. Mobile backhaul assessment The service lifecycle can be divided into four phases: network construction, service turn-up and burn-in, service assurance and service troubleshooting. Network construction describes the physical wiring of the network elements. Service turn-up relates to validating that the Ethernet service has been installed and provisioned correctly. Service assurance relates to measuring the key performance indicators (KPIs) for 24/7/365 service, validating that service-level agreements (SLAs) are being met, and ensuring that any service problems are being detected. Service troubleshooting relates to fixing problems that are detected and validating that the fix actually solved the problem. Construction Before turning up new services, technicians must characterize the physical link that connects the tower to the network. In most legacy networks, the physical link is made of copper. In new 3G/4G deployments, fiber is becoming the preferred medium. Characterizing the link allows a field technician to locate physical-layer faults, qualify the copper local loop, determine link loss and return loss measurement, qualify WDM channel and validate connector cleanliness. Turn-up and burn-in Service turn-up and burn-in is the only time operators can perform full-line-rate testing; it is essential to test and validate that services (i.e., VLAN and CoS) are properly configured/provisioned to deliver the performance defined in a customer’s SLA. Typically, this includes a guaranteed level of throughput, a guaranteed level of burstable throughput as well as a guarantee on loss, latency and delay variation, etc. These tests can be saved for future reference. Comprehensive reporting is essential to demonstrate that the service delivered meets the SLA. Performance criteria need to be met: network availability and mean time to repair (MTTR) can easily be verified, but performance criteria are more difficult to prove. Demonstrating important parameters (e.g., performance availability, transmission delay and delay variation, link burstability and service integrity) cannot be done with a single ping test. Testing a network’s QoS requires tools to simulate real-life scenarios, including multistream traffic generation, per-flow analysis and measurement of QoS parameters. The International Telecommunications Union (ITU) has recently standardized a new service activation methodology – Y.1564 – which is geared toward Carrier Ethernet and takes into account the important KPIs of the Ethernet services running over today’s network architecture. The ITU-T Y.1564 document describes a two-phase approach to measuring the performance of Ethernet-based services over point-to-point or point-to-multipoint networks. The first phase is the service configuration test, which verifies that the network is correctly configured for each service. The second phase is the service performance test, which provides a method to test the long-term quality of the network. EtherSAM (Y.1564) offers an ideal test strategy for today’s and tomorrow’s wireless backhaul networks, while reducing cost and improving MTTR. Service turn-up can be accomplished in one of two ways: inside-out testing or outside-in testing. Inside-out testing refers to running a turn-up test from a central test head to a device at the demarcation location that is in loopback. The device at the demarcation location can be a cell site router, a NID, a handheld, or a performance endpoint device, as long as it is standards-based. Outside-in testing refers to running the turn-up test from the demarcation location to a central test head. Inside-out testing tends to be more economical than outside-in testing because it does not require another person at the demarcation location. By automating the procedures necessary to put the remote device in loopback, inside-out testing can be run whenever necessary. One important aspect is to capture key service metrics in both directions to record any direction-specific characteristics. Before mission-critical applications are fully deployed, a 24-to-72-hour testing period is required. This serves to evaluate the service performance in real-life conditions. Operators can remotely or locally verify the performance of backhaul connections with line-rate test traffic by running the EtherSAM (Y.1564) or RFC 2544, and other nonintrusive testing protocols, such as the Ethernet OAM CFM suite (IEEE 802.1ag and ITU-T Y.1731) – without requiring a disruption of the traffic flow. Service monitoring Wireless service providers are not only facing OPEX challenges as their backhaul networks grow exponentially; they also face new QoS and QoE challenges. To acquire and retain customers, operators must focus on their subscribers’ QoE and use it as a competitive advantage. Continuous monitoring requires a centralized and integrated service assurance solution that automatically launches periodic tests from an MSC (mobile switching centre) to all MSC-served cell sites through CESR (carrier Ethernet switch/routers) and/or performance endpoints to allow operators to address possible problems without deploying unnecessary resources. Performance endpoint devices (PEPs) provide wire-speed loopback capabilities, integrated responder capabilities (including two-way active measurement protocol support) and OAM capabilities for non-intrusive, in-band testing (including support for IEEE 802.1ag and Y.1731 OAM). Tests can occur from a centralized test point to the PEP or from a portable device located at a specific test point to the PEP. Ideally, PEPs are deployed at all handoff points to provide remote, in-service test capabilities. By doing so, operators can perform continuous monitoring and ensure that any degradation is quickly detected. Because the device is always located at this point, there is no need to send technicians to the test site if or when an issue is detected. Technicians can troubleshoot the segments that lead to the path location issues and determine which segment is failing. Troubleshooting Troubleshooting begins when anomalies are detected during the monitoring phase. Real-time actionable information enables operators to remotely launch performance metrics, improving the MTTR (mean time to repair) by segmenting the network and isolating problems. Unfortunately, the monitoring tool cannot always pinpoint the exact cause of a network fault. When additional testing is needed, the tools used during the installation and commissioning can be used to find intermittent or hard-to-diagnose problems and provide the ‘birth certificate’ that can be used has a performance baseline. IEEE 802.1ag (CFM) provides many of the tools necessary for segmenting service problems. This standard defines mechanisms that allow a provider to test to various key points. Also, continuous testing to these key points in the service path in addition to the demarcation location provides valuable historical information that can be used to identify hard-to-debug, transient problems. The biggest advantage of the CFM tests is that these can be run without having to take the traffic down. For even further troubleshooting, one or more turn-up subtests (throughput, latency, loss or back to back) can be used. The key aspect of these service-troubleshooting procedures is to restore service performance and avoid truck rolls as much as possible. For advanced troubleshooting requirements, multi-stream traffic can be generated at wire speed to the performance endpoint or a handheld tester. Mobile operators must now evolve from only managing network performance to also managing service performance. Testing the network with a simple ping is not enough; performance indicators must constantly be validated and measured on a per-service basis. Understanding the mobile backhaul network architecture is very important. During installation, commissioning, and turn-up, one must validate the SLA parameters to certify the quality of the services being deployed. This certification enables providers to create ‘birth certificates’ and then use them as a report for their end-customers. PEPs are the perfect addition to any network: they are the right tool, designed to fulfill the need for test responder and loopback capabilities; they are cost effective and enable scalability. Using a 24/7 monitoring system helps providers ensuring that their services meet SLAs. If the monitoring systems have testing capabilities, they can provide a first level of troubleshooting without any truck rolls. To investigate further and pinpoint the source of the problem, the same portable test equipment used for installation is perfect. Test equipment and monitoring systems become the stepping stone to better QoS and QoE, while providing greater OPEX.