Network slicing is a primary feature of Standalone 5G but E2E orchestration of the network and services presents challenges
Operators are facing challenges along multiple vectors with the complementary yet distinct transitions to Standalone 5G, disaggregated network systems, and cloud-native operations. At the same time, there are multiple vectors of opportunity to leverage network investments into new service revenue streams. A key piece of aligning the reality of 5G with the hype it has generated is E2E orchestration of both the network and of services, according to Alla Goldner, director of technology, strategy and standardization with Amdocs.
“This all together into the same picture actually brings you the critical technology enablement from one side and, from the other side, a demand for those new on-demand services,” Goldner said in an interview. “Orchestration, design, inventory and assurance are really four key elements of the E2E orchestration solution.”
The North Star here—encompassing the shift to 5G Standalone, the flexibility that comes with multi-vendor network systems, and increasingly automated cloud-native operations—is the ability to spin up a network slice. This cross-domain virtual partition of a network would provide the user with exacting service parameters on-demand while enabling the operator to make the most efficient use of network, spectral and other resources; and (ideally) it would all happen automatically. But, as Goldner notes, standing up a slice that spans the core, transport, RAN and edge domains is a challenge.
“By now,” she said, “everyone realizes that there is no single killer application. The key of 5G is actually in combining those applications” and delivering them on-demand. “Building such a slice across a different domain or supporting several applications at the time by allocating those slices is definitely a technological challenge and quite complex and quite advanced network management capabilities are required here.”
Network slicing requires a holistic view of network domains and service assurance
As operators work towards network slicing and attendant capabilities, Goldner reiterated that there’s no single orchestration solution that manages everything, rather it’s a layered approach. “The highest layer would be service orchestration which supports super complex service decomposition into the network, translating parameters of the service into network parameters.”
She likened this layer of orchestration as providing an overview of domain-level orchestrators for the core, transport, RAN and edge. The highest layer “gets and correlates information from those domains, and manages those domains in a closed loop. That layer has the most complexity…and also that layer basically is the layer responsible for instantiating, for establishing, the slices.”
With a service orchestrator talking to domain orchestrators, the next need is for unified design of slices according to service demands “which works E2E by getting into different domains and design parameters within those domains,” Goldner said. Next is inventory “which is close to a real-time single source of truth that covers both network and service parameters and sees everything which is happening to the network in real time.” Last “but not least” is running service assurance which itself also has a layered architecture. Assembling these layers in a hierarchical, logical way allows operators to conduct E2E orchestration of network slices and the services traversing those slices.
In addition to the E2E orchestration piece, another best practice is to develop what are essentially network slice templates modeled on service parameters and necessary network functions that may be shared or dedicated to specific network slices. Further, Goldner noted that operators need to be aware of complexities that come along with integrating telecom networks with centralized and distributed, and public, private, and/or hybrid cloud infrastructure.
With regard to multi-cloud operations and management, “There is actually a big technological challenge of how you combine those clouds, how you manage dynamically the service, and deciding finally how you instantiate those network network functions and then those slices.