Are you over-engineering you mobile backhaul networks? Here are five reasons why this may be happening:
1. Legacy backhaul technology
Many operators have now moved away from backhaul based on multiple T1/E1s, or even Ethernet over Sonet/SDH, and toward IP/Ethernet. But those operators that still have such technologies in place are constrained by inelastic bandwidth levels. Only a limited set of rates can be supported and the effort associated with upgrades is very large, encouraging over-provisioning.
2. Poorly instrumented networks
Many networks are not sufficiently instrumented with Service OAM, such as Y.1731 or TWAMP. As a result, operators cannot adequately measure critical user experience-related parameters such as delay, delay variation and frame loss across the network. At the same time, most mobile networks sample actual utilization levels every five or 15 minutes. Such intervals are far too long to give an accurate view of utilization given the bursty nature of packet networks. With limited network data at their disposal, operators prefer to create cautious forecasts.
3. Inaccurate network topology and inventory
Even where an operator is using modern IP backhaul in a well-instrumented network, inaccurate network and topology inventory can cause problems. Running complex performance and utilization analytics against the wrong topology data results in untrustworthy information on where network upgrades are needed.
4. Analytics that do not scale
A large operator network with tens of thousands of cell sites will generate massive amounts of utilization and performance data – in effect, "big data" – for analysis. The algorithms that process this data and enable operators to create network upgrade plans are complex, especially as they must cope with network changes in real time. Most legacy operations support systems (OSSs) have not been designed to handle big data, nor are they underpinned with big data technologies. Operators do not have the analytical tools they need for a more surgical approach to capacity planning.
5. Limited automation of the upgrade process
Even if operators had instrumentation and analytics to tell them exactly what to upgrade when, most do not have the resources to upgrade capacity at individual cell sites more frequently than once in two or three years. In most operators, the upgrade process for a cell site is a complex multi-step process. It involves ordering additional bandwidth from access providers (for off-net locations) and tracking this to fulfillment, provisioning changes on aggregation and cell site routers, turn-up testing of new or augmented backhaul circuits and rehoming of services where physical capacity of resources is exhausted. Upgrading all the sites of a 50,000-site network over three years means upgrading almost 1,400 sites every month – a highly resource-intensive activity. Without significant automation improvements, it would be impossible for operators to upgrade more frequently – for example, once a year – as they simply would not have the headcount to do it.
This blog post is an excerpt from the CENX & Light Reading white paper written by Caroline Chappell, “Aligning User Experience & Cost in Mobile Networks Through Just-In-Time Capacity Management”.
To learn more about tuning your network with just-in-time capacity management, watch our free webinar "Achieving a Finely-tuned Network Through Just-In-Time Capacity Management" on demand.