Why adoptions of virtual BNGs are limited?
We see strong demand on virtualizing Mobile data Core systems like EPC however there is not so strong drive in virtualizing fixed data systems like BNG, why is that, and what is the industry best practice?
Telcoss are heavily investing in virtualisation/containerisation across globe. However their current focus area is virtualising/containerising mobile network functions as the volumes are exponentially higher compared to fixed line consumers, e.g. in emerging market the fixed line broadband subscribers are around 1% of the overall mobile consumers, hence replacing existing low volume physical nodes by VNFs/CNFs not necessarily result significant business benefits.
Hence the reason for stronger push currently on virtualization for EPC is likely that for most of the telcos more revenue is coming from Mobile, and not as much from fixed line business.
The need is not enough to for a virtual BNG in emerging markets. If we look at Jio’s stats, the FTTH subscribers are 1% of their mobile subscriber bases. For household data requirements, there was some mention about “On average, each 1000 subscribers generates about 2.5 Gbps of incoming and 0.5 Gbps of outgoing traffic..” ( https://vasexperts.com/blog/bng/licence-vs-subscription-for-vbng/) and at that rate there will be <5000 license/nodes required.
There is another factor affecting investment around Virtual BNG is Mobile/Fixedline convergence due to 5G. Operators would need to make a decision how they converge Fixed/Mobile and that may potentially decide the investments around VBNG (https://www.broadband-forum.org/marketing/download/MR-464.pdf).
From technical stand point, the VBNG architecture across NEPs are not matured enough to provide programmability, scalability and it’s performances are yet to be established. For example, an vBNG appliance with x-series h/w would still require 2.5 to 3 times increase in compute when data plane requirements increase by 4 times. Operators like Vodafone, who are pushing for disaggregated multivendor architecture, are possibly investing in Virtual BNG and they have tested with CASA, Benu etc., however an article suggests economy of scale, efficiencies are yet to be proven (https://www.telecomtv.com/content/open-networking/vodafone-takes-its-open-zeal-to-the-broadband-access-network-42502/).
Deutsche Telecom has rolled out its Access4.0 framework with multiple SDN vendors for SDN enabled broadband access, however it has been done for a handful of customers (https://www.telecomtv.com/content/open-networking/deutsche-telekom-switches-on-its-disaggregated-broadband-network-40635/).
Given number of telcos are getting into FTTX business in recetnt times, nother aspects to be looked upon as how recent the investment is in equipments and have they gone for vendor locked in equipmnets and does vendor have a roadmap to adopt standardised architecture (e.g. Open APIs). Depending on the same, telcos may need to wait before they emabark on on a complete revamping and switch over from appliance based BNGs to virtualized ones.
So below are the points for Telcos to consider:
- Growth roadmap for their fixed line business
- Roadmap and timeline to move to optical fiber from copper
- End of life of current equipment
- Maturity of Virtual BNG architecture and provider (openness, scalability, programmability, performance)
As SDN (Software Defined Network) strives to virtualize most aspects of networking, BNGs will eventually be part of the same. The timing will be dependent on revenue driving prioritization and when there will be economies of scale.
