In a complex, multivendor network environment that is taking initial steps toward virtualization, the ability to have real-time insight into network conditions is becoming increasingly important.
“There’s a split coming between the fast data analytics that are required for network control and optimization – real-time functions – and the big data analytics that are deeper and allow you to get into customer experience management and understand the trends in your network and things that would help marketing departments do a little bit more targeted marketing and get ROI there,” said Scott Sumner, VP of solutions development at Accedian Networks.
Interoperability is a real challenge in getting to uniform network visibility, Sumner said, and the polling that is often done to gather data typically doesn’t happen fast enough to be useful to operators. He also said that as network operators move toward more virtualization, concerns about performance and the accuracy of measurements can hinder effective migrations to virtualized network functions. In addition, there will be a wider range of places in the network that need to be instrumented in order to get visibility that is no longer centered on hardware. The sheer amount of data, he added, means that automation and machine learning will be key.
“Things are happening on a per-service, per-application, potentially per-user basis in real time, and the … answer to that is you need analytics, but you also need to use some deep learning or machine learning to be able to take that network performance visibility that’s happening and correlate it with the subscriber state or the subscriber’s context, … the time of day, how the network is configured and come up with some sort of proactive or pre-emptive way to solve problems before, or as they’re occurring, and report the necessary correlations,” Sumner said.
Watch the full network visibility discussion with Sumner below:
[embedyt] http://www.youtube.com/watch?v=si7H1jus6Pk[/embedyt]
Another question is how to approach big data projects in telecom. Anssi Tauriainen, director of business development for analytics with EXFO, said that EXFO’s approach is in terms of service assurance. He said the company often sees a technical focus in big data projects, looking at the capabilities of the platform – how fast it is, how many queries can be made, how much data can be captured, which can actually result in a loss of project focus. As long as a platform is scalable and flexible, he said, starting with the most critical use cases and then being able to change and expand them as time goes on provides more benefit than trying to do everything for all users at once.
“It’s easy to lose your focus … trying to build something which fulfills every single user’s needs for the analytics,” he said.
While it is necessary to have a structure or platform set up to collect data, he said, specialist solutions can be added which provide the granularity and specific data mining that are needed by users across the service provider organization, from customer experience management to radio network optimization and network planning. See the interview with Tauriainen below:
[embedyt] http://www.youtube.com/watch?v=K1mGuKFofhY[/embedyt]
Want to learn more about the uses of real-time analytics and big data? Read RCR’s special report and watch the webinar.
Follow me on Twitter! @khillrcr
