Machine learning
Machine learning
There are two distinct node type, storage nodes and bandwidth nodes, each being a network peer that the corresponding provider type operates in order to provision their service to the network. There is a reference software implementation of each node, called Colossus and Argus, respectively, but in principle there could be alternative node implementations for the same underlying protocol described in document.
Node
Storage nodes are primarily involved in
- accepting uploads from users
- downloading data to be stored from other storage nodes,
- uploading data to other storage nodes and bandwidth nodes that may require it,
- dropping data which is deleted from the network,
and bandwidth nodes are primarily involved in
- uploading data to users upon request,
- downloading data from storage providers in accordance with local caching policy,
Service-level agreement
There is no direct protocol level enforcement of what service-level agreement each node should conform to, for example in terms of:
- storage capacity
- up-time
- up-speed
- connection capacity
- latency w.r.t. a given location
Want to learn more?
For a detailed overview of how each node works, please review the: Nodes overview
Related Terms
Machine learningCreator TokenCreatorConsumer appsNFTS