Limits of locally managed mesh
A new class of Wi-Fi products form a mesh network to coordinate with each other to increase the Wi-Fi range. Current mesh routing protocols are designed to provide reachability of traffic between mesh nodes, only ensuring that the traffic makes it to the internet gateway in some way. This focus of mesh routing on the survivability of the backhaul traffic between mesh nodes largely ignores the routing demands of the connected devices. These unsophisticated routing protocols do not address application performance or wireless network capacity to achieve the desired customer QoE. In fact,
most mesh systems available today operate on a single channel backbone thereby significantly limiting the overall capacity while being prone to self-generated interference.
Additionally, the locally managed traditional distributed control plane mesh routing architecture increases the complexity of each individual node making it difficult to continue to add capabilities by adding additional nodes and routes to the mesh system.
Advantages of a centralized cloud based approach
The proliferation of Wi-Fi nodes, or pods, throughout the home provides large degrees of freedom for traffic routing between the end-device and internet gateway connection.
The number of potential connections between nodes increases by N(N-1), greatly increasing the ability to deliver a reliable, high-performance Wi-Fi service. Distributed Wi-Fi networks are therefore more complex to configure and manage, specifically to deliver the optimal performance commensurate with their capability.
This complexity is best handled with a centralized software entity such as Plume's cloud with knowledge across the entire network. In essence, a cloud based approach achieves a superior wireless system by shifting the complexity from hardware to software.
Other advantages include:
- Assigning devices to pods (Client Steering) is more effective with a global view of the entire network, ensuring the best route.
- Feature, update roll-out is simpler, faster and less risky than always needing to update firmware.
- Algorithms calculate the optimizations in the cloud and then it configures the network all at once instead of several independent devices, each running separate calculations with uncoordinated timing to choose and apply a new configuration.
- A cloud-based management system is able to aggregate data from many homes for continuous analysis and learning. Improved methods for network optimization, client device behaviors and bugs, typical device/user patterns and behaviors can be extracted from such a cloud-based centralized database.
- The compute, storage, and memory complexity of each individual node is reduced, making the nodes smaller, less power-hungry and easier to develop and deploy. The cloud, with virtually unlimited compute power and memory, can run arbitrarily complex algorithms to learn and optimize.
- Complete visibility and control of the network from anywhere with an internet connection. This allows customer control anywhere through the app and better visibility and control for support personnel.