In the opening salvo to his recent Whitepaper called “Demystifying the Role of Next Generation Network Planning,” Heavy Reading analyst Ari Banerjee said that for service providers to succeed in the hyper-competitive communications marketplace, they need to first know their subscribers, understand how they use their services, be able to deploy innovative business models that maximize revenue, and be able to efficiently manage network resources.
The first two requirements should not be that hard. Service providers have access to all the data they need to make those determinations about customers and their usage. Telecom companies get criticized for not meeting the third requirement – deploying innovative business models – but connecting practically the entire planet with voice, broadband, wireless and data services and making billions of dollars doing it is pretty innovative. Granted there is more they can and should be doing in response to new competitors and new service delivery models, but that’s not where their biggest problem lies. It lies with the fourth requirement: efficiently managing network resources.
The key to achieving this, Banerjee said, is operational excellence. This is no excuse, but except for a few shining examples like Amazon, Apple, FedEx and others, every company struggles with operational excellence and even these models of excellence experience drift occasionally. However, service providers have a pretty serious case of drift. They need a new gimbal.
It won’t be easy getting centered again when the demand for access to existing network resources doubles 18 times over the next five years as the prevailing Cisco model predicts, yet projected spending on support infrastructure increases by only 10 percent, according to a new Amdocs study. Getting on and staying on the track of operational excellence will take some serious planning, literally. Planning tools are key to enabling the kind of excellence winning will require.
It won’t just take better planning, it will take a new kind of planning tool, Banerjee said. Today planning is still primarily an offline tool that lacks the dynamic quality necessary to plan next-generation networks.
“You need a tool that can plan and predict, optimize, prioritize and anticipate ahead of the demand curve,” Banerjee said. “It is based on anticipation, but it is also based on real data from real underlying systems. It is based on analytics.”
Planning isn’t just about capacity management and resource allocation. Planning tools need more data to be able to incorporate something new into the planning process: customer data, device data, usage data and application data, inventory data, deep-packet inspection data, data from switches, routers and other network feeds, and get this – revenue data. Banerjee said planning has to bridge the gaps between business and operational support systems with capacity-management systems, a familiar theme as we move into 4G and beyond.
Most service providers (65 percent of those surveyed) still plan to use analytics-driven planning for operational planning, but more are looking to use it in other ways, such as real-time service assurance (62 percent), price and product mix optimization (57 percent). In addition, Banerjee says network planning has to be more revenue-focused and connect network elements to revenue-generating initiatives.
“Having subscriber and plan revenue relate to network capacity can be an extremely valuable business intelligence tool for marketing and finance executives,” Banerjee said, adding that it can provide valuable input to capital allocation during tactical capacity planning. This will help finance teams better understand the profitability of the network and even distinct portions of the network.
Another new trick planners have to add to their bag is the ability to deal with virtual elements in the network. Banerjee said there is a major requirement for operators to optimize the deployment of virtual nodes and that service providers need to adopt a more accurate and timely approach to capacity management and that will require the principles of advanced analytics to be applied to network planning.
Following the Path
One vendor Banerjee mentioned going down this path is VPIsystems. Once a tradition planning and engineering firm, VPIsystems now delivers an analytics-based optimization and capacity-planning solution called OnePlan Predictive Analytics that includes system configuration, user administration, database management and logging and reporting. It manages multiple analysis or planning scenarios across network, IT and business, and provides specific and quantitative answers to the decision-making processes associated with modeling new services and applications, as well as the new capacity required to support them.
The analytics lets OnePlan model multiple scenarios across multiple technologies and planning environments. Its focus is to solve business challenges for service providers migrating from 3G to LTE, not just around network elements but also their interaction with different types of subscribers, devices, services or any combination of them.
VPIsystems’ Russ Green
Russ Green, senior vice president of marketing and business development at VPIsystems, says the big difference between a traditional planning system and one that is analytics-based is that the latter mo longer speaks the language of erlangs and dropped calls – but speaks at a more business level.
“We don’t come up with network demands for capacity planners in the context of megabytes per second or erlangs because it doesn’t translate into marketing speak,” Green said.
The business is driven by business parameters and if you want to engage all departments to participate in the planning process, you have to speak their language. “It makes for a much more meaningful planning mechanism,” he added.
Network-performance data is still key to planning, Green said, the difference is that with analytics you can either use that data to model scenarios to derive the best capacity planning picture or you can use it more as business intelligence to determine how much revenue is coming in from a particular part of the network or from particular devices or applications.
“Having a dollar value associated underlying traffic helps you understand which pieces of the network are producing money and by adding a little bit of information from the BSS you can tell how subscribers are performing,” Green said.
One of the challenges to analytics is the variety of data formats and sources of data. Some scenarios such as with unlimited flat-rate services are pretty simple stuff, as the data from call-detail records is fairly standard; however, the plans with which providers try to differentiate themselves are not very standard and throw a wrench into the data.
“That’s why we have a very flexible extract, transform and load tool which can take structured data from anywhere,” Green said.
With these multiple sources of data, service providers can tell not just which parts of the network are most profitable, but which parts are being consumed by which type of subscriber. Unfortunately, even with better planning tools, the one thing the tools can’t deliver to service providers is time. And there never seems to be enough time to take the time to plan.
“It’s a rush,” Green said. “It’s like the Oklahoma land grab out there and the real problem is there are no extra staff in most cases to properly manage a rollout.”
For two reasons, some service providers may have more cushion than others in taking the time to plan their 4G rollouts. The providers with the deep pockets got started first, but also paid the price of being first to market and had to figure things out as they went. This isn’t always ideal, but Green said despite the concerns across the industry over capacity, most providers have sized their initial rollouts properly and have included a lot of cushion for data volume growth. However, he said, nobody got the signaling demands right and signaling traffic has become one of the greatest concerns with LTE.
“LTE is a flatter network and there is nowhere near the integration complexity of 2G and 3G, but with that there is a lot of signaling across the packet core all the time. It isn’t hidden behind switches or the radio network controller anymore,” Green said. “So I don’t think there is enough time being put into planning. Part of that is because people don’t really know where the problems are yet.”
Although they do know there is a problem with signaling. Outages at Verizon and in Japan’s LTE network were due to signaling issues.