The cable industry has been actively launching various initiatives such as cloud computing, WiFi, and the connected home over the past few years. However, in the pursuit of additional profits, the fundamental technologies that support these revenue-generating services can become afterthoughts. Three crucial issues should be at the forefront in the push to increase average revenue per user:
- Tackling the growth in bandwidth consumption
- Managing multiple device types
- Dealing with the complexity of IPv6
I’ll look at each of these issues in detail over the next few blog posts. Let’s start with the first one — tackling bandwidth consumption.
The growing popularity of over-the-top (OTT) subscriber services such as streaming video, cloud-based applications, and video conferencing results in additional bandwidth demand. This creates bottlenecks during peak hours and can deteriorate the user’s experience. Last year, peak hour Internet traffic increased by 41%, while average traffic grew by 34%. Peak hour traffic will continue to outpace average traffic over the next several years, causing even more infrastructure strain.
Today, subscribers find even temporary service blips unacceptable. “Bill-shock”, caused by unexpected over-consumption charges, is also a leading point of contention between customers and providers. A holistic picture of how and where bandwidth is used within a home network can educate customers on what and where charges are stemming from. Thankfully, a variety of technologies are available to help you keep tabs on service quality, network health, and subscriber consumption patterns.
Simple Network Management Protocol (SNMP) provides vital information about underlying network usage. This polling mechanism has been used in the wireline and cable industries for decades and allows for the extraction of data from most networking equipment. This includes quantitative measures, such as the number of bytes or packets sent or received through any interface on the network, as well as qualitative data, such as the number of dropped packets or retransmission. Time-based SNMP data analysis allows you to find out where performance problems may be occurring.
Another technology widely used for network analysis is IP Detail Record (IPDR) collection. This protocol enables an application to gather usage patterns and allows you to collect, process, and manage IPDR records from cable modems, embedded multimedia terminal adapters (EMTA), and set-top boxes through routing elements (such as CMTS). Information includes data sent and received through specific interfaces, as well as service flows required by certain applications like video streaming, voice, and social media. Unlike SNMP, IPDR uses a “push” mechanism and imposes less strain on the network.
When you correlate SNMP with IPDR data, you can obtain a holistic view of network health. You can then cross-reference this bandwidth consumption information with network data to identify where and when bottlenecks are occurring, what applications and services are hogging bandwidth, and who the higher users are in a particular region or during peak periods.
SNMP and IPDR are not new technologies, and to be useful, you need to gather data over long periods of time. With the advent of Big Data, it is now feasible and desirable to store this information to perform analytics on the network.
When choosing a SNMP or IPDR tool, consider these factors:
- Avoid network strain (push vs. poll)
- Obtain a clear picture of system health
- Map data to actual subscribers
The last thing you need when trying to overcome issues of overconsumption is to cause more network strain. Unfortunately, some tools rely on active polling to gather network intelligence, so make sure you look for tools that are efficient and scalable.
Stay on top of your network’s health by selecting a tool that will collect historical data and also alert you to current network health issues. Streamline your data collection and reporting process through one interface and provide specific details such as corrected and uncorrectable errors, utilization, signal-to-noise ratios, etc.
Finally, while overall consumption data is useful to identify traffic trends and aid network planning, this data must be easily mapped to specific subscribers so that they too can gain an understanding of how they are using bandwidth. Look for a tool that easily integrates into your OSS, which can categorize traffic into different services flows.