Customer experience key for network performance monitoring

Castle said “historically the way engineers have tested, they’d go around with phones, they’d ask ‘can you hear me now?’, they’d drive around, which is an interesting way to test your network, but not a great way to understand what is happening to your customers”. Now, the company is making use of social media and other tools to understand the user experience.

“The nice thing about social media is that people will tell you what they think. The good, the bad, the pluses, the minuses, and they don’t pull any punches. If you want to get better, you have to listen, especially the bad stuff.”

He added speed test apps are also an important tool, because “customers are trying to tell you something, they are literally testing your network for you. And what’s nice is that it’s super easy to get that data, and we get each data point – what phone it is, what location it was, what time it was. We have hundreds of thousands of customers doing free engineering testing. That’s amazing”.

Castle was less positive about benchmark tests, because “they hit a limited geography in a limited amount of time. You can only drive so many places, you can’t drive in anybody’s house, but they take this data set and somehow say ‘that’s how these networks perform’”. He claimed such tests also lack device diversity, and “not all devices are created equal”.

In contrast, crowdsourced data covers all times of day, across geographies, indoors and outdoors. “That’s a hugely powerful data set,” he said.

Serge Marokhovsky, director of product management for telco service assurance at VMware, highlighted how network analytics and management have been impacted by various levels of added complexity.

“It’s not just the technology that has been evolving and got more complicated, it’s the use cases. As new generations of technology come into use, new products, new services come out – I don’t think we can foresee some of the future things that will come with 5G,” he said.

With this complexity, the amount of data generated by networks skyrockets. That inevitably means automation, Marokhovsky said.

“Automation really becomes necessary because it’s humanly impossible to manage, to administer all of this. Even if you are in China and manpower is a lot cheaper, when you are trying to handle hundreds of millions, if not billions, of events on a daily basis, it’s not humanly impossible.”

Anand Srinivas, CTO of enterprise-oriented firm Nyansa, said: “The application perspective is important, because at the end of the day, the only purpose of a network is really for client devices to access applications, and do so in a positive way.”