When going to cloud-native 5G, forget what you know about telecoms: Dish exec Said Berrahil
U.S. operator Dish is in process of building out a cloud-native 5G network that adheres to Open RAN principles and specifications. Extending from this idea of a cloud-native network, the company identifies itself as a cloud company that’s providing a telecommunications network as a service. But what does that mean, what does it entail and how do you do it?
Dish Vice President, 5G Network, Said Berrahil talked through the vision and execution at the recent Open RAN Global Forum. He said the company’s goal of operating telecom workloads in a public cloud has been the subject of significant industry pushback, but Dish has done it and is now focused on turning exposed network data into autonomous network operations.
“For me, this is only the beginning of the journey,” Berrahil said. “Our journey is to launch the world’s first cloud-native, data-centric, fully-automated network of networks.” And this means breaking down network domain silos and creating a “cloud continuum” linking the radio access, core and transport networks. “This is unique because, from one base station to another base station, it’s the cloud…I don’t see a difference between the edge, the core, the RAN, the transport. It’s all in the cloud for me. And from that point of view, that’s what I meant by the telecom becomes just a feature that is on top of the cloud. We’re not a telco company. We’re a cloud company that offers a telco service.”
Berrahil noted the lowering capex and opex is only one aspect of building a cloud-native network. The real value, he said, is in bringing the innovation speed of cloud into the world of telecoms. In order to effectively innovate, data has to be generated and actioned; he described this as not just testing to validate but testing to monitor and learn.
Automated testing sets the stage for automated operations in a cloud-native 5G network
By taking a cloud-native approach to 5G, Dish is able to a new level of network data observability. Berrahil explained that testing is foundational to the company’s goal of network automation and is focused on adapting an IT model of “active testing…They’re capable of correlating [active and passive testing] to do some predictive analysis and monitor their network…We want to bring this all to telco. But then, there is a next step and this is what brings us to Dish, is that we don’t test only to validate. We don’t test only to monitor, but we test to learn. The number one reason for me for testing is to provide enough data for our data products, our AI machinery, to exist. There is no data science without data—bit of a simplistic view but I need to have enough data for me to be automated and to have AI and to be predictive. And this is why we have testing-as-a-service.”
Berrahil said his goal is to run 10 million test cases per day—right now, he said, he’s off by a few zeroes but making progress. And the need arises because the level of network complexity means ongoing operations and, as a function of that, testing, has to be a machine-led activity. “We are looking for behaviors and we are injecting noise in the network, provoking the network, pushing the network to its limit in order to learn from it. And this is done in an autonomous manner.”
Data is fed into an orchestration layer where decisions are made. “I would like to stress,” Berrahil said, “…this network cannot be managed by humans.”
What does a cloud-native 5G, data-centric approach mean for the wireless workforce?
While the network isn’t intended to be operated by humans and many vital operational processes, testing for instance, are becoming automated, there’s still a need for specialized workers. As Berrahil explained, when he started with Dish he prepared to recruit radio engineers, core engineers, etc…He said his boss told him, “‘You can only recruit data scientists.’ And I was like, ‘Why?’”
“It’s just data,” Berrahil said. “It is not RAN. It is not core. It is not transport. It’s just data. This has been my biggest discovery.” Data scientists, he said, don’t particularly care what data they’re looking at–epidemiological, associated with elections, the weather–”They don’t care. So long as they have data, they look for patterns.”
In addition to data scientists, Berrahil said Dish also emphasizes recruiting developers. In practice, the company organizes “Bonded Operational Agile Teams,” or BOATs, that are stood up to solve specific problems and led by captains. “Once they define the scope of the team, of the work that they want to do, they go from at berth to underway…And they do it for two, three, four, five, six months until the product is either delivered or we stop doing it because there is an element of failure in it.” To watch the full interview with Berrahil, as well as all of the content from the Open RAN Global Forum, register for the on-demand virtual event.
