Meter’s method is totally different. All {hardware} runs an identical firmware. Telemetry information has the identical construction whether or not it comes from an entry level, change or firewall. This consistency lets Meter practice its Command AI mannequin on a clear dataset.
Command now operates in two areas: help and operations. For help tickets, Command analyzes real-time telemetry, runs mannequin jobs and recommends actions. The system at the moment handles 85% of tickets with model-generated insights. Meter reserves 15% for management group testing.
For operations, Command automates community design. After gross sales calls the place prospects describe their setup, Command reads transcripts and generates full community configurations. This consists of topology, structure, required {hardware}, VLAN assignments, IP addressing and safety insurance policies.
Desired state configuration takes flight
Conventional community configuration is mostly crucial. Engineers log into switches and routers to run instructions that change gadget state. Every configuration change requires CLI entry or API calls to particular person units. Configuration drift occurs when units get out of sync.
“All the supply of reality and intelligence really lies within the again finish, and that is what we name desired state networking,” Varanasi stated.
The structure builds on software-defined networking (SDN) ideas which were available in the market for over a decade and abstracts them out additional. The supply of reality lives within the cloud, not on particular person units. Engineers outline what the community ought to do. Meter’s system figures out easy methods to configure {hardware} to attain that state.
