Vertical interactions Sample Clauses

Vertical interactions. The vertical (north/south) interaction introduces new concepts such as aggregation and hierarchical scaling into the picture. A fluid domain can be created by a FLUIDOS “supernode” that aggregates multiple nodes, hence exporting a virtual space that is the union of the resources (and services, objects) of the composing nodes. The supernode-backed fluid domain is created with the same mechanisms already in place for the horizontal interactions; individual nodes are still in control of their own resources (ownership principle) and can withdraw them (if allowed by the signed contract). In addition, a special “tethered mode”, available upon explicit configuration, allows individual nodes to be completely controlled by the supernode, giving up their local intelligence (and thus autonomy and ownership) in exchange for a simplified control plane. This mechanism, for instance, would be used to bring resource-constrained devices, such as micro- controllers, into the FLUIDOS ecosystem. The aggregation paradigm can be applied recursively, with “supernodes” becoming part of a bigger “hyper-fluid domain”, managed by an “hypernode” (or controlled, if the “tethered mode” is enabled), for a theoretically endless number of hierarchical levels. This provides the foundation for the hierarchical scaling property of FLUIDOS (recursive hierarchical architecture, from a single device to domains of domains). In this respect, the north/south interface, which enables interactions with different nodes (either local devices, clusters or other FLUIDOS “supernodes”) all featuring the same interface, will be backed by novel resource aggregation algorithms that can aggregate the resources of each single node (normal, super, etc) in a scalable way. Finally, a FLUIDOS Broker is an aggregation, consolidation, and brokering (i.e., reselling) point that supports also interactions between multiple administrative domains, whose duties include the capability to (1) observe the resources offered by many domains and the associated performance when involved in task offloading; (2) to monitor (and predict) the quality of network connections; (3) to suggest the “best node” when asked for an offloading request. However, given that super/hyper nodes include brokering activities albeit limited within the boundaries of a single admin domain, FLUIDOS will reuse for the above nodes the great part of algorithms (and software) developed for the Broker, assigning to this component a major role in the architec...