Understanding network optimisation
16 May 2011 | Guy Matthews
Network optimisation isn’t just about troubleshooting. It allows carriers to understand how their bandwidth is being used and optimise their assets. Guy Matthews investigates.
There has never exactly been a time when network performance wasn’t important. But in an era when operators of WAN infrastructure are expected to also be application specialists, services can be hosted hundreds of miles from their point of use and international transactions need to complete in milliseconds, the performance of most networks is under tighter scrutiny than ever.
With carriers making less and less margin out of basic transport, they are having to provide other services where network performance is under a more granular level of scrutiny, says Eric Bear, director of product management at Visual Network Systems, a vendor specialising in network testing and measurement equipment for enterprise and carrier customers.
“To get more revenue, they are extending into buildings and into the delivery of LAN and cloud services,” he says. “They’re extending right into the data centre and so need an appropriate set of tools that allow for better visibility and for problem fixing. Applications running slow? They need to fix that. They need to fix things fast and all while keeping costs down.” The complex web of wholesale relationships makes this job harder, he believes: “There are so many partners and suppliers to take account of,”says Bear. “Plus there’s the migration from traditional interfaces like Frame Relay and ATM to Ethernet and IP.”
Accelerating the network
The era of cloud and virtualisation necessitates that a carrier needs not only visibility of the whole network, but of everything running on it right down to the level of the application, he argues: “You need new toolsets that operate inside the virtualised environment,” he says. “We provide such performance management tools for people like AT&T, Verizon, Sprint and NTT. They can then build them into their SLA calculations, so that their customers can understand how bandwidth is being used, as well as for their own operational troubleshooting. If there’s a problem with a network-to-network interface (NNI), you need to be able to point quickly to which carrier is responsible for it so it can be solved.”
The most basic objective of evaluating and testing any network will always be to at least match the performance offered by competitors, argues Owen Cole, technical director of vendor F5 Networks, which has worked with many carriers on accelerating traffic. “There’s an element of disquiet and insecurity among service providers,” he says. “They don’t want to be thought of as just a pipe. They’re asking how they can differentiate and add value. Accelerating the network is a hot topic, coming in different guises but with a similar end objective.”
Changing market pressures are reflected, he says, in the dialogue that F5 is having with carrier customers: “We’re talking to them as always about acceleration, but now with a bit more on top. They want to do more than their competitors, more than just making the network go faster. They want to make the actual apps run faster, making the experience of their customers better. We come at the optimisation market from the apps end.”
Variable speed limits
Nigel Pink is managing director of Ipanema, a developer of network optimisation and WAN governance technology which it sells to service providers like BT, Cable & Wireless and Orange, as well as large enterprises. He believes that there has been rapid evolution in the type of optimisation products being supplied to carriers and service providers, to match their changing needs. “You started off with vendors of basic acceleration solutions, putting a box at either end of the network,” he explains. “That became a dirty word, and people started talking about network optimisation. Imagine a Ferrari on the M25 with no other traffic. Well, in reality the M25 is not free of traffic and does have a speed limit. We help people to manage the variable speed limits. We help carrier customers to win business, and face the cloud and the content explosion. As a carrier, you want to be confident about the performance of critical applications.”
Improving network performance is not always a simple matter of adding bandwidth, he believes: “More bandwidth is a tempting solution, but it won’t always fix the problem,” he says. “You need to govern what’s going on – WAN governance is what we call it. Most CIOs don’t know what their infrastructure is costing them. Imagine that sort of ignorance applied to another industry - like trucking.”
As well as pressure to optimise their own network, carriers face demands from enterprise and wholesale customers to help them optimise their WAN assets too. “We talk to enterprises active in maybe 30 countries, facing the possibility of at least 30 different network operator relationships,” said Chris Kimm, VP of sales engineering at Verizon Business. “They come to us to make sure the network issue doesn’t get in the way. They want us to transform their network into a workable state. It’s not always just a matter of ‘turn up the bandwidth’; it may be more about prioritising one type of traffic over another. To achieve this we use a variety of WAN optimisation devices, from vendors like Cisco and Juniper. It gives us and other carriers the chance to provide value and deliver what the customer really needs. It works for our wholesale business too. In negotiations between carriers on routing traffic, you’ve got to get the right quality of service so a service provider can offer the right solution at the right price.”
Enabling the network
Ian Smith, a consultant with Orange Business Services, reports evolving needs among the company’s enterprise customer base: “The meaning of the term ‘optimisation’ has changed a little in recent years,” he says. “For us now, it’s about enhancing performance for customers, enabling their business to perform better from the various locations where people are sited. Our remit is extending though. Customers want to know how to centralise applications in one location, but now it’s evolving more towards understanding how those apps can perform better, and setting SLAs against that. Customers are smart about what network performance is acceptable and what to expect these days. They don’t want someone to put in an optimisation box and walk away. It’s all about understanding what’s on the network before any box gets near. The bar has risen, and is rising all the time.”
Mid-sized enterprises now want optimal network performance, not just major multinationals, says Bertrand Meis, CEO at vendor company OneAccess: “We sell our solutions to telcos like Orange Business who sell them to enterprises,” he says. “We do multi-service routers, as well as Ethernet demarcation devices. In Ethernet, we operate between the traditional Layer 3 market, driven by players like Cisco and Juniper, and Layer 2. We’ll deliver a single device that manages both layers in one – the first time the two have converged. On the router side, we enable telcos to go out and be champions of cloud and software-as-a-service (SaaS). They have the network, but we enable it for this world. The technology has existed for large companies, and now we’re doing it for mid-sized too.”
These days the pressure is on to optimise network performance at every level, access as much as backbone, says Floyd Wagoner, director of global product marketing at Motorola: “Optimisation for us means lowering the cost of deploying fibre to meet tomorrow’s need,” he said. “Copper is dead and FTTH is the future, and we optimise the cost of its installation.”
The optimisation industry has been forced to come of age, believes Smith. “Five years ago, reporting on network performance was atrocious,” he claims. “Now customers want in-depth analysis: they are asking what? where? when? why? Vendors of optimisation solutions are leading the way – well some of them anyway. Companies like Ipanema are helping to increase the granularity of the information available. We can firefight problems better, or let customers do it for themselves.”
He believes that there is some way to go before optimisation amounts to a concise deliverable, complete with standards: “Some vendors are old school – unable to give us a roadmap to take to a customer,” he concludes. “If we’re going to keep on delivering more for less, no one-stop solution does everything. Optimisation is very complex, and likely to remain so.”
“Network optimisation methodologies will become more sophisticated,” predicts Jan Tjurin, head of network planning and optimisation for Nokia Siemens Networks. “Besides network-centric optimisation KPIs, like dropped call rate and handover failures, subscriber-centric indicators are being introduced to measure and improve service quality. This will give service providers insight into the quality of their various services and optimise the network where it benefits the largest number of subscribers or provides the best return on investment. A structured approach to network optimisation will likely become the norm. This, coupled with a degree of automation, will replace many of today’s ad hoc optimisation arrangements. The development and roll-out of LTE is providing a catalyst for this trend. Lastly, I foresee that self-optimising networks will deliver a high degree of automation and reduce operational expenditure, cutting out the need for human intervention in network optimisation processes.”