Telstra Connects Its Network Plan With Its Market Plan

As Telstra began developing its strategy for a vast next-generation network transformation, it recognized that the advent of new, bandwidth-hungry services like video and massive multi-player gaming (MMPG) would require a different approach to network planning. These new, large and immersive applications would require a more integrated and “market-based management” approach. Traditionally, network planning was based on trends and estimates of future demand, or in response to trouble spots. Disconnects between engineering and marketing created issues when it came time to launch new services and expand the network accordingly. Telstra decided to use network planning software with a new set of processes that would enable it to build its networks and add capacity in direct alignment with customer demand, thereby making network planning more proactive and capital-efficient.

Traditional Network Planning: Follow the Network
The network planner’s job has always been technically challenging, but also well-defined. The planning organization typically interacts with the CTO, CFO and network operations groups. With the CTO organization, planning analyzes new service requirements, determines the capital needed to meet anticipated network demands, and sets engineering guidelines for technology and implementation. Armed with this information, the planning organization then works with the CFO to estimate the capital expenditures (CAPEX) required over the next several years, given the anticipated network demands.

The planning organization works finally with the network operations department to plan major new network additions and projects. These additions may be due to known new users or the introduction of new technologies or services. Equipment is allocated according to the engineering guidelines, and a first-cut network plan is handed off to the engineering department. These interactions enable the planning organization to plan the outside plant, switching, and metro and long-haul transmission domains.

The traditional TDM network is simple, in the sense that when a resource is allocated to a service, it is “used up”—no longer available to other services. To determine whether resources are needed in a given transmission link, switch or multiplexer, it’s only necessary to monitor thresholds. Typically, warnings start at about 85 percent, with criticality considered about 95 percent. If a piece of equipment hits 85 percent capacity and historical data suggest it will reach the critical stage shortly thereafter, it can be flagged as an area requiring allocation of more equipment.

Budgeting for the network used to be just a matter of modifying the prior year’s plan, constrained by the financial requirements of the upcoming year. Planning departments traditionally would order as much equipment as they could within the budgetary constraints, and allocate it as well as possible using spreadsheets, manual methods, rules of thumb and help from equipment vendors.

Telstra’s New Approach to Planning
Telstra realized that with IP, the rules of the game had changed. The notion of network capacity being “used up” is not applicable in the next-generation IP network, where over-subscription is possible and dynamic resource allocation happens more automatically. A more sophisticated planning approach would be necessary.

For years, some have argued that the easy answer to this problem is overbuilding. But Telstra wanted to avoid the “30 percent rule” dictating that networks be overbuilt by 30 percent over the theoretically lowest cost network. This approach involves pre-building hot spots months in advance, and it results in wasted capital, due to network overcapacity, on the order of 10 percent.

Telstra also knew that its old planning methods could add another 10 percent of overcapacity in trying to plan for the repercussions of major outages caused by natural catastrophes, and another 10 percent on top of that to account for the methods equipment vendors would use to meet Telstra’s network requirements. At this point, it became relatively clear that better informed and more frequent network planning could reduce or eliminate at least 20 percent of new network expenditure.

The company had already been in preliminary conversations with VPI Systems regarding its OnePlan application and noted that a cross-domain approach that used a common planning tool would make sense. Its conversations with BT and its major technology suppliers and SIs, which had evaluated the system already, confirmed this and Telstra chose to move forward.

BT is working with VPIsystems to integrate its marketing, business planning and engineering intelligence and has completed pilots and initial implementations in the transport and access domains. Focused on its 21CN initiative, BT is developing “best current practices” for network planning in its converged, IP-based networks including processes for data loading; network capacity planning; and connecting network design into its existing processes. At the end of this ongoing, multi-phase project, BT will be able to plan, optimize, acquire and deploy network capacity across the entire 21CN and will practice “demand-aligned” capacity management going forward.

Telstra initially planned to implement its network planning solution as a follow-up to its new inventory project, as was the norm. When it was recognized, however, that it could build out predictive network plans before spending any capital dollars, the planning solution, which now supports more than 250 of Telstra’s next-gen network designers, was fast-tracked ahead of the inventory program.

Converged IP-Centric Networking
Telstra’s new network architecture is based on IP and Ethernet standards and uses technologies like MPLS to ensure proper quality of service. Networks based on these technologies tend to route and re-route traffic automatically and are far more difficult to plan. Their shared-fabric and self-routing characteristics require planners to have systems that determine how the network will behave under normal, overloaded and failure scenarios.

IP will drop or delay packets during overload conditions and introduce new impairments like digital jitter and delay that affect interactive services such as voice, two-way video and gaming. Network planners therefore need to understand how services will be affected under varying network conditions. In addition, they need to know how the network can be configured to provide the best service at the least cost.

The issue is further compounded because the interaction between network layers creates significant complexity. For example, Telstra’s new services will ride on the IP and/or Ethernet network, and high-QoS services will use special routes, but all of this traffic rides on the underlying logical transport network—ring or mesh—which in turn rides on the underlying optical infrastructure.

For the planning teams, the effect of traffic on each layer must be taken into account in the other layers. Planning reliability and disaster scenarios becomes very complex as a result, because backup resources must be available at each layer in the hierarchy (for example, at the IP layer, the SDH layer and the underlying optical layer).

The New Model: Market-Driven Planning
Trending has always been central to network planning, but its value is greatly diminished going forward. Major new projects need to lay out new network architectures, but there’s no but historical data to analyze. Similarly, new applications demand and share more bandwidth, but there are few if any large scale examples to use. Third, Telstra is decommissioning legacy infrastructure, but trending can’t do much to anticipate the impacts of network turnover.

Rather than a trending based approach, Telstra would drive its master plan from market information, as opposed to network information. Traditionally, marketing had been treated as more of a back-office function than network operations. This mind-set had to change. Marketing had to be moved out front, performing market projections for new services based on geography, related demographics and time.

With marketing setting these projections, the network resources that are needed to support marketing’s plan can be determined, including the type traffic, varying QoS requirements and the underlying equipment required by each service. This view gives Telstra an idea of what network resources are required to support market demand in each geography and according to specific time periods, such as six-, 12- and 24-month cycles.

With this total projection of required resources available, existing and already planned network resources can be subtracted. What remains is the additional required network capacity per service, per geography and per technology. Added to that picture are the requirements for the next-generation network build-out and any planned legacy network decommissioning. This aspect of the new planning process requires interfaces to the systems, primarily inventory, from which the new planning application can derive a view of the current network and near-term planned network capacity additions and subtractions.

Planners then perform additional what-if analyses to determine network resilience to overload and disaster conditions, adding in additional network capacity as dictated by business policy. Using the new integrated planning system, the network planners can lay out the new network additions using the standard equipment configurations. This information passes to the financial and logistics teams for ordering the right equipment at the right time, and to the operations people for the physical layout of the network additions, organized by projects.

To tackle implementing this planning system, which manages a complex process across multiple network domains and business organizations, Telstra adopted a phased approach. The first phase has consisted of collecting market information by service; planning the access network, which is receiving most of the financial investment for next-generation networks; and determining the underlying optical transport requirements. The subsequent half-dozen phases will bring in the Ethernet, IP/MPLS, VoIP and IMS planning, as well as legacy equipment decommissioning.

Trending, in this new process, becomes a check on the market-driven plans. The planners look at trending information, by link and equipment utilization, and ensure that the market-driven master plan lines up with what trending reveals. In the slow-changing, overbuilt environments common earlier in this decade, the traditional six-, 12- and 24-month plans were reasonable. Now carriers realize that this frequency is not sufficient. They are planning to cycle through entire network master planning scenarios at least every month, if not every two weeks. This can enable them to adjust next-generation network rollout plans as they learn more about how to perform them right. It also allows them to monitor how the market responds to new services, and to stay ahead of network requirements as demand grows for new and bandwidth-hungry services.