SPAN is a fully distributed network architecture. That is to say, routing, processing and storage are distributed to every single device in the network. Every device is capable of caching, processing and switching content and reports its capability and status to the global intelligence. This enables unconstrained local and global decisions for the optimised placement and processing of content. The network becomes maximally efficient. The distinction between server, cloud, and network loses any significance. The network is the cloud and the computer.
Virtual & elastic
SPAN network functions are abstracted from hardware. They are provisioned as containerised applications running on specialised or generalised hardware such as proprietary network switches or Wintel/ARM Linux servers and open routers/switches. We are working with major network equipment vendors to build SPAN stacks alongside, and interoperable with, their legacy TCP/IP stacks. SPAN Virtual Network Functions (SPAN-VNF) are managed by the SPAN Management Framework (SPAN-MF). SPAN-VNF are modelled on highly sophisticated cloud net flow edge models, enabling the provisioning of network services anywhere in the network. Once again, the network is maximally efficient.
SPAN uses an interoperable combination of traditional TCP/IP addressing and name based addressing and routing. A content name request such as SPAN://rhett/bondi/disney/batman/directors_cut/4K/HDR is self-certifying and self-explanatory. It enables switches to route requests and to build Pending Interest Tables (PITs) and Satisfied Interest Tables (SITs). Addition of tags such as hashes allows distributed storage using existing schemes such as blockchain, Protocol Labs IPFS, Merkle trees, etc. Machine Learning (see below) enables switches to make intelligent switching and caching decisions. SPAN is a statistically balanced protocol. Admission control and loop prevention are inherent, as are broadcast and multi-cast. Publishers and distributors publish once to the network with QoS parameters. The network takes care of storage and distribution.
SPAN-AI agents are independent and self-governing. Machine Learning (AI) capability allows local optimisation of all network and operational functions. Agent swarms use algorithms inspired by nature. Agents feed capability and real time status up to the global intelligence (see below).
SPAN-AI agents use Machine Learning to locally optimise network content placement and processing decisions based on local network conditions. Global training and optimisation pipelines optimise agent and global performance. This optimisation is based on algorithms developed over 10 years at Bell Laboratories and refined in real time by the SPAN-MF (management framework).