The introduction of IP/Ethernet solutions into mobile backhaul networks gives operators not only new opportunities but also new challenges as they need to validate and assess the performance and the quality of their services. Vikas Arora, CTO of EXFO, tells Keith Dyer how using the right test and service assurance solutions will be crucial to the success of operators’ backhaul migration strategies.
Mobile Europe: Vikas, what are the overall priorities of your customers, the network operators and backhaul network providers, as they consider migrating their legacy mobile backhaul infrastructure to IP/Ethernet?
Vikas Arora: The key thing is cost reduction. Currently backhaul is predominantly carried over T1/E1 lines and that represents the single largest area of OPEX spending operators have within their networks. We tend to hear a fairly constant number from operators — that somewhere around 25-30% of their overall OPEX goes into the network operations, with backhaul forming a significant part of that.
So it’s not a secret that operators are looking to reduce the cost of mobile backhaul, and as a part of that they are looking to move to a more scalable and cost-efficient technology like Ethernet to support future bandwidth requirements and to protect future investments.
Operators also know that most outages originate from the mobile backhaul network, with backhaul accounting for 40-60% of problems, compared to 15% due to the base station and just 5% in the RF. So alongside the drive to lower OPEX, operators also need to increase the reliability and availability of their backhaul networks.
Mobile Europe: Yet as they do that, they are faced with renewed challenges in terms of assuring and validating services, which are encapsulated within Ethernet in a very different manner from TDM networks.
Vikas Arora: Yes, IP/Ethernet presents operators with a totally different environment from TDM, meaning that operators need to move from just testing network performance, which was acceptable within TDM, to testing network and service performance within Ethernet.
Additionally, despite the fact that operators are introducing this new networking technology, they know they can’t afford to take too much time to stabilise those networks, and they also cannot have the technology take too high a financial toll on their methods and procedures. That means they need to look at whether they have the ability and the tools to manage the user experience across the network in a centralised and scalable manner, whichever architecture they have chosen.
So in terms of operators’ requirements, as they move to packet-based IP/Ethernet backhaul networks, they need to make sure that they have three things. First, access to the right tools to help them manage this new environment. Second, they need to train field and operations staff to ensure their expertise and thinking evolves from TDM to IP/Ethernet. Thirdly, they need to update their service turn-up test and troubleshooting procedures to align them with the IP/Ethernet service activation requirements.
Mobile Europe: You have mentioned that operators will need new capabilities to be able to manage the introduction of this new service environment. What are the tools they will require to ensure SLAs across their IP/Ethernet networks?
Vikas Arora: Well, I think it pays to step back a bit and have a look at the full network life cycle, from network construction and service turn-up, through the “burn-in” period, and then to operational service monitoring and trouble shooting.
Some predictions are that backhaul bandwidth requirements per tower will grow from 10Mbps in 2009 to 100Mbps by 2014. To deal with this exponential growth in data demand, we are seeing the deployment of more fibre to towers. As operators drive this fibre to the towers, they then need the right tools to be able to characterise the fibre and validate the link performance, to ensure that IP/Ethernet services can be delivered properly.
The next phase is the service turn-up and burn-in. In this phase, with the transition from T1/E1 based backhaul to Ethernet, operators need to test and troubleshoot a lot more parameters than they test when turning up T1/E1 circuits. This is the “totally different environment” I referred to earlier.
Let’s look at an example. For T1/E1 connections, service turn-up involved testing the Bit Error Rates (BER) and round trip delay measurements. Yet for Ethernet-based backhaul services the tests required are much more detailed. The parameters operators need to consider in this environment are, first of all, measuring throughput, frame delay (latency), and burstability. These elements are all critical for interactive voice, video and many data applications. The time sensitive nature of non-buffered video, and other data applications such as VoIP, also means that the measurement of frame delay variation, or jitter, is vital.
Then there are frame loss measurements. Frame loss ratio is crucial for voice and video quality and the capability to measure it over time will provide an idea of the long-term integrity of the service. You cannot do this with a simple “ping” test of the network.
Finally, Ethernet brings with it the need for VLAN and Class of Service (CoS) validation. Each Ethernet virtual circuit (EVC) will support more than one class of service to handle different types of IP services, so the testing and validation needs to be done for each VLAN and class of service simultaneously.
This means that to assure quality of service within an Ethernet environment, operators need the tools to be able to, first, validate the network configuration for each defined service (traffic shaping, rate limiting, etc) and, secondly, to validate Quality of Service KPIs and prove SLA conformance. The output of this phase is a service “birth-certificate” with all the service parameters tested and the SLA signed off.
The next phase is to move beyond that initial birth certificate into long-term, post turn-up operational performance and SLA monitoring and troubleshooting. This phase deals with the need to provide 24×7 measurement of KPIs, alerts and alarms on service degradation, and aggregation and analysis of KPIs.
Mobile Europe: I understand there is some movement in Ethernet testing methodologies, to be able to provide service activation and ongoing service assurance.
Vikas Arora: That is a very good point. RFC 2544 is the current Ethernet service turn-up testing methodology yet RFC 2544 was designed for network device testing in the lab and not for service testing in the field. It does not include all the required measurements for today’s service SLA parameters, like jitter, out-of-sequence, QoS measurements for multiple concurrent service levels, etc. Moreover it provides a snapshot view, and does not adapt to long term measurements. As it is a sequential test methodology, it is also very time consuming.
To address these limitations and to reduce the operators’ OPEX, we, working with number of other industry players, have proposed a new Ethernet testing recommendation called ITU-T Y.156sam (service activation methodology).
For on-going service assurance, OAM standards play a critical role. The OAM standards, 802.3ah and 802.1ag/Y.1731, are just being adopted by the equipment vendors. 802.3ah deals with the link fault management whereas 802.1ag/y.1731 deals with connectivity fault management and performance monitoring. It provides us with connectivity check messages to monitor the heartbeat of the network, loopback messages to determine where a fault is, and link trace messages to find the affected path. All these checks need to be done for each VLAN/CoS within an EVC, so it’s clear that operators need a scalable solution which leverages these standards and provides an ongoing service assurance system, with the right technology support, to monitor SLAs and rapidly reduce the time-to-fix in troubleshooting.
Mobile Europe: Where is the industry in terms of ITU-T Y.156sam adoption?
Vikas Arora: It’s still early days, but because Y.156sam addresses operators’ needs to speed up service turn-up and reduce OPEX, we are seeing widespread support. There have been some very good contributions from the industry to make sure the Y.156sam recommendations address all the operators’ key needs and that this standardised approach becomes part of operators’ service turn-up methods and procedures. This standard is currently in draft form and is expected to be ratified by early 2011.
Mobile Europe: So you’ve built up a picture of why operators must consider a new approach to their rollout of Ethernet backhaul, allowing them to have optimised turn-up procedures and ongoing performance monitoring for multiple classes of service in a cost-efficient manner. How would you sum up EXFO’s role in meeting those goals?
Vikas Arora: There are two clear messages we want to deliver. First, at EXFO we look at the full network life-cycle and are delivering the right tools and solutions with the right level of automation and standards support. We are also playing a leading role in the standard bodies to help evolve the methods and procedures which address Ethernet mobile backhaul needs as operators move from one phase to another.
Secondly, to improve operational efficiency and productivity, we are working with operators to provide solutions which help capture and store field test, troubleshooting and service turn-up birth-certificate data right through the life-cycle so that this can be correlated with the service performance data collected during the operational performance monitoring of the Ethernet service.
Our end goal is to deliver a future-proof, standards-based, solution to the operators to help reduce OPEX, improve productivity and deliver operational efficiency and actionable intelligence.