There always are multiple ways to solve a capacity or function problem using a communications network, as these are engineering decisions, and all such decisions involve trade offs. No one solution always works for every problem.
In that regard, there clearly are capacity limitations on early 5G network slicing networks. It is not possible, at the moment, to support unlimited virtual private networks. The issue is “how many” can be supported.
The alternatives become more valuable as the ability to support a network slice diminishes. If the objective of the slice is to reduce latency, ensure quality of service or bandwidth, then edge computing, premises computing, dedicated capacity and redundant connectivity links all are alternative solutions.
Right now, if asked how many slices can be supported, the answer is that it depends. Even when use cases might be local, such as many or all devices on a campus, capacity of the core network is invoked. In principle, could a slice be localized? Yes. That is essentially what private 5G networks do. But if so, is a network slice actually needed?
And if the objective is latency performance, in many instances an SD-WAN, leased private capacity or an owned transport network or edge computing could suffice. In other cases, where the performance must be supplied on a highly-distributed basis, such as to support gamers at many distinct locations, a network slice might be the best solution.
No comments:
Post a Comment