Page 303 - Cloud computing: From paradigm to operation

Framework and requirements for cloud computing 1

– allow updating the resource information across multiple CSPs in synchronization with the events
(e.g., reserve or release of resources) involving the CSPs;
– allow periodically, or on a request basis, collecting information about the usage and performance
status of the resources of multiple CSPs;
– allow periodically, or on a request basis, collecting information about the resource availability
(e.g., dead or alive status of machines) of multiple CSPs;
– allow exchange of monitoring information in commonly defined ways across multiple CSPs.

9.3 Resource performance estimation and selection
The resource performance estimation and selection capability deals with the selection of resources from the
candidate resources that have already been reserved in peer CSPs. This capability estimates the achievable
performance of available reserved resources and assists the CSP in the selection of resources to be effectively
used.
The resource performance estimation and selection capability is recommended to:
– allow estimating the achievable performance of available reserved resources (e.g., computing
resources, storage resources, input/output capacity between storage resources, network
bandwidth) in the secondary CSPs.

9.4 Resource discovery and reservation
The resource discovery and reservation capability deals with search, discovery and reservation of available
resources in the peer CSPs. This capability also deals with reservation acknowledgement for the candidate
resources that have been tentatively reserved in the peer CSPs.
The resource discovery and reservation capability is recommended to:
– enable discovery of resources available in the peer CSPs;
– allow the reservation of discovered resources in the peer CSPs;

– allow provisional reservation of discovered resources, i.e., to keep the resources to be used (as
candidates), for later acknowledgement (for some of them) or release (for others);
– allow finding available resources in the peer CSPs based on different priorities (e.g., in a different
order of searching);
NOTE 1 – Quality requirements may vary from service to service and each resource contribution to the service
quality may vary as well. For example, if latency is critical, it should be possible to first reserve resources in
the servers that are near to the user and then reserve the network resources. In contrast, if bandwidth is
critical, it should be possible to first reserve resources of the networks that can provide sufficient bandwidth
and then search for available resource in the servers that are connected to those networks.
– allow reservation of available resources in the peer CSPs on the basis of different priorities
(e.g., early recovery, required quality guarantee, service type, etc.).
NOTE 2 – For example, a vast quantity of resources is required for recovery from a large-scale disaster.
However, all required resources may not necessarily be available. In that case, it should be possible to
forcefully reserve resources for lifeline services rather than for other services.

9.5 Resource set-up and activation

The resource set-up and activation capability deals with the set up and activation of reserved resources in
the peer CSPs. This includes connecting to the peer CSPs via networks, remotely activating (i.e., invoking)
software and transferring or copying data to enable the use of resources in the peer CSPs.

The resource set-up and activation capability is recommended to:
– allow the establishment of reserved resources in a peer CSP;
– allow accessing to the configuration and policy settings of reserved resources in the peer CSPs.

295