New network traffic patterns created by Covid-19 have demonstrated the shortcomings in traditional network capacity planning. Provisioning for several ‘busy hours’ is no longer sufficient and requires too much human input. In the future, networks will need to rely on automation and continuous network adjustments. CSPs will need to rely much less on human monitoring and intervention, and much more on automation and the cloud. This is needed to ensure capacity and quality can be maintained.
All operators have reported sharp increases in voice, video, messaging, and basic Internet connectivity as a result of Covid-19. In order to address these challenges, content providers and network operators have had to engage in throttling and voluntary reductions in video resolution from HD to SD in order to preserve bandwidth. However, all of these measures point to the need to build future networks with an eye toward resilience and flexibility to address the next major crisis.
Overprovisioning network capacity
Traditionally, network operators have used ‘overprovisioning’ as the primary means of capacity planning. This involves determining the minimum capacity required to address ‘busy-hour’ traffic – the times during which network demands are highest – and then planning for an amount above that amount (e.g., 20% higher than the expected demand). This process is actually quite complicated. It requires sophisticated algorithms that take into account performance characteristics of multiple network components to estimate ‘supply’ (capacity) and a similar level of rigor to estimate user behaviors, taking into account things such as localized events (e.g., stadium events), changes in demand based on weather, and the like.
Flexible network architectures
The calculations used to estimate overprovisioning are still valid. However, traditional network management is being pushed to its natural limit currently. Traditional ‘busy hours’ are being replaced by multiple pockets of activity and network demand throughout the day. This is based on the effects of a significant proportion of the world’s population sheltering at home. Simply put, the new paradigm lays bare the shortcomings of the traditional over-capacity model. As a result, one of the enduring effects of the current pandemic should be to accelerate the already ongoing movement toward more flexible network architectures. In these, capacity and quality can be monitored and adjusted in real time, with minimal human intervention.
Real time monitoring for performance
Unfortunately, while many operators are working toward making their networks more flexible and capable of spinning up capacity in real time based on shifting demand patterns, this movement is still in the early stages. Most operators still operate static network environments, planning long-term, very high-granularity capacity upgrades only when necessary. However, with network performance becoming even more critical in the future, operators will need to monitor and control their network in real time. This is needed to ensure performance of critical services and, ideally, deploy software-activated capacity dynamically in response to traffic pattern changes.
The current crisis should make clear that building networks for resilience and flexibility is becoming a national security issue. Operators should use the current situation as an inspiration to up their game. They need to start re-architecting their networks and network control mechanisms. This will ensure smooth operation of services in all foreseeable scenarios, including widespread emergencies such as Covid-19. National governments should do everything in their power to encourage this process.