Previously, we looked at some of the data in Heavy Reading’s survey of network operators’ plans for disaggregation conducted in December 2020. Respondents had to work for a verifiable network operator and be involved in network planning and/or purchasing network equipment. There were 82 qualified participants that were counted in the results. Volta was pleased to be a sponsor of this project and a full copy of the report by Heavy Reading’s Sterling Perrin is available here. In addition, the results were the basis of a webinar (available on-demand) which included presentations by the operators Telefonica and MTN.
In that post, we pointed out some of the responses that showed how operators view the importance of disaggregation. Let’s look at the responses on RAN splits and virtualization which are major drivers for DCSG.
Heavy Reading asked operators to estimate the 5G RAN functional splits options they expect to deploy in their emerging 5G RANs between 2020 and 2023. The results had to add up to 100% and showed operators expect to utilize a range of options:
“While there is no single standout, the results, nonetheless, are a strong endorsement for split architectures generally. While the traditional distributed macro architecture will account for roughly a one-quarter share of emerging 5G based on the results, nearly two-thirds (63%) are expected to be some form of partial or full centralization involving functional splits. Heavy Reading finds this result to be a strong endorsement of centralized/split architectures and a departure from Heavy Reading surveys in previous years in which the majority of respondents favored traditional macro architectures for 5G.”
An overwhelming majority of 93% of respondents anticipate some level of RAN function virtualization. Among the options, virtualization of both CU (layer 3) and DU (layer 2) functions scored the highest, selected by 48% of survey takers. At a distant second, 28% of respondents are interested in CU virtualization only in locations where DU functions are also virtualized.
Heavy Reading asked survey respondents to identify where they expect their network edge computing functions to physically reside:
“At 56%, central office locations topped the list of preferences, but the surprising finding is that all locations are in play – from the cloud data center to the cell site to the customer premises. Each location was selected by at least 40% of operators.”
Implications for Networks
Heavy Reading noted that disaggregation allows operators to source their switching hardware and their routing software from separate suppliers. This approach to disaggregation can eliminate vendor lock-in, increase agility in responding to network change, speeding the time to bring innovation to market, while reducing capex – which were operators most important stated goals in adopting disaggregation.
“This approach has become particularly important in IP routing. The DCSG segment, promoted by TIP, is one prominent example of the hardware and software disaggregation model, but the NOS/hardware separation model extends to other segments as well, including edge and core routing.”
We will wrap up our review of the survey in our next post with a look at operational issues and how they impact TCO.