Page 251 - Cloud computing: From paradigm to operation

Framework and requirements for cloud computing 1

Appendix II

Use cases for configuration of distributed cloud
(This appendix does not form an integral part of this Recommendation.)

This appendix identifies use cases for configuration of distributed cloud.

II.1 Autonomous cloud service provisioning on distributed cloud

Table II.1 – Autonomous cloud service provisioning on distributed cloud

Use case
Name Autonomous cloud service provisioning on distributed cloud.

Abstract This use case describes the autonomous feature for provisioning a cloud service on
distributed cloud.
The distributed cloud supports dynamic service provisioning by utilizing core cloud (CC),
regional cloud (RC) and edge cloud (EC). EC mainly supports low latency to provide a cloud
service. For EC, it is sometimes hard to provide the cloud service due to various reasons (e.g.,
limited capacity of EC, initially requested service, energy saving). In the case where EC is not
ready to provide the corresponding cloud service, EC is required to make a decision to
provide the service to the CSC as rapidly as possible. For fast decisions required by time
sensitive services in EC, a request to a centralized point (e.g., core cloud and regional cloud)
is not appropriate due to network traversal time; the request should be handled in the EC
itself. To support autonomous cloud service provisioning on EC, associated context
information is required. Associated context information is necessary to be updated on EC
periodically and aperiodically. To avoid synchronization burden, associated context
information is summarized and updated (e.g., calculating average, grouping value, choosing
representative value). To avoid frequent transfer data to the core/regional cloud by edge
cloud, the edge cloud filters and deliver to prevent noise from multiple end devices and
deliver batched data to core or regional cloud.
In conclusion, autonomous provisioning is a main feature that differentiates conventional
cloud and distributed cloud. Through the feature, CC, RC and EC on distributed cloud are
loosely coupled. The feature also supports fast cloud service provisioning on the distributed
cloud. Autonomous provisioning is done by associated context information updated
periodically and aperiodically.
NOTE – Examples of decision: 1) Deploying the cloud service and provide the cloud service
on the EC, 2) Delivering the request to neighbour EC, 3) Delivering the request to the CC, etc.

243