Fast access to network data is key to successfully migrating to IP networks. Clearly, improved infrastructure intelligence lets operators more effectively detect and measure network and application performance, as well as the customer quality of the experience. However, with current packet acquisition methods, service data from the network isn’t adequate to let providers see and control a network effectively. Information from the core is abundant, but data from the edge isn’t, and the result is an overall network intelligence deficit.

With the number of Internet-connected devices predicted to reach 50 billion by 2020 and mobile traffic expected to increase 18-fold by 2016, mobile service providers face significant challenges to maintain a high quality of service for customers while increasing revenue.  As wireless operators migrate to modern networks that are handling significantly higher bandwidth and traffic, improved network intelligence moves from being a competitive advantage to becoming mandatory for efficient management and adequate quality of service.

Gaining access to data at the edge of the network, where the majority of end user problems occur, is a significant challenge that current solutions network service providers have available to them do not address. What’s needed is a new paradigm redefining customer, content and network intelligence. This redefinition will require a new method for gathering raw packet data from a distributed Ethernet network.

Improving network data gathering in networks depends on intelligent packet acquisition. Intelligent packet acquisition promises to change the game by providing dramatically improved visibility while minimizing the cost of access. Smart Network Application Platforms, for example, provide ubiquitous, real-time access to critical network data, replacing traditional methods that can’t scale to the growing number of connected devices.   

Current Packet Acquisition Limitations

The methods and procedures used to access packets in an IP network have evolved over time, but the basic premise of deploying appliances at points of interest has remained the same.  However, today’s networks are far more complex with challenges such as dynamically rerouted traffic, priority queues and class of services that change from device to device and link to link. These challenges exponentially increase with end-user bandwidth demands and real-time applications and services that are very sensitive to jitter and loss. Network intelligence and quality-determining decisions need to be made with data from the edge of the network close to the real customer experience.   

Traditional packet acquisition approaches have relied on instrumenting a network with switch port analyzer/mirror ports, taps, and in-line network appliances to feed data collocated to an analysis application or tool.  However, these approaches continue to have significant drawbacks.

The high cost of accessing data  — Proactively installing purpose-built or dedicated probes and analysis applications everywhere visibility is desired may be practical in the core of the network, where there are reasonable numbers of high-speed links. However, deployments near the edge quickly become prohibitively expensive because of the number of access points rises exponentially.  

High mean-time-to-data  — Due to the high cost of dedicated taps and appliances, many critical network elements are simply not instrumented.  This means when a service affecting issue occurs, a truck roll is required to gather data from a remote site and this means knowing where to gather the data in advance and it often means gathering data after the event.

Lack of uniform and pervasive access throughout the network — Modern day networks use different elements from different network equipment manufacturers, all with differing capabilities and features.  This means operational inefficiencies and data inconsistencies. Providing consistent, uniform and pervasive access to solve complex problems in a multi-vendor, multi-element network poses a significant challenge. 

What Does a Viable Solution Look Like?

If you can decouple the data collection and filtering from the management, aggregation and analysis and then distribute throughout a cloud-based network, you can dramatically reduce the cost, footprint and complexity of capturing rich intelligence about the network, the content and your customers’ experience. Remote visibility all the way to the network edge – where 80 percent of system impairment issues occur – and using a cloud-based approach to separate data collection and filtering from management and analysis is vital.  This provides unprecedented access to intelligence across the network and the following benefits:

Affordable access to packets — Acquisition must be part of a standard network component widely distributed at the network edge. Embedding intelligent packet acquisition using an integrated circuit in an optical transceiver quickly extends the data reach to virtually any optical link anywhere within a network. 

Dramatically lower operational expenses — With an intelligent transceiver replacing a standard transceiver, no additional rack space or external power is required. It doesn’t need an internal operating system or external configuration and maintenance. Using the host network for communications, no management or overlay networks are needed.

On-demand access — Combining this type of packet access with a multi-user, multi-probe access platform means technicians do not need direct access to network elements to quickly and easily access the targeted data they need. Users will view any subscriber, service, or quality of experience on demand, leading to quick diagnoses and faster problem solving.

Ubiquitous visibility everywhere in the network — Again, this approach can be deployed into any pluggable transceiver port throughout a network. This increased visibility to problems results in reduced mean time to fix problems, which in turn leads to increased customer satisfaction.

Customer Intelligence – The data captured for network monitoring can be extended to many different applications providing analytics and intelligence on customer usage and behavior.In conclusion, packet acquisition solutions will dramatically improve network performance on virtually all levels. It will also deliver savings in capital and operating expenditures through open architecture, modular design and time-synchronized captures. Operators will centrally locate analysis tools and expert personnel, making them more efficient and cost effective. Most importantly, it will preserve the value of existing infrastructure and personnel while simultaneously enabling these resources to perform better and more efficiently.

Doug Fantuzzi is vice president and general manager in JDSU’s Communications Test and Measurement business segment. Before joining JDSU, he held a number of executive positions at Amdocs, Telcordia Technologies and Bell Labs, supporting software development and helping to launch OSS products.