Page 309 - Cloud computing: From paradigm to operation
P. 309
Framework and requirements for cloud computing 1
Table I.3 – Inter-cloud use case: Performance guarantee regarding delay
Use case
– Network connections among interworking CSPs are instantaneously
established or reconfigured. Then service-related data including user
identifier (ID), user data and application data are transferred from the
original CSP to the CSP that is leasing the resources.
– Access from CSCs is appropriately changed to the interworking CSP so as
to achieve route optimization and thus mitigate the performance
degradation caused by the distance from the original CSP.
– As a result, the CSC, who keeps the same user ID, can continuously
access the service at the same level of response time as before.
Information flow – Relevant CSPs are supposed to join a common trusted alliance
(federation) in advance and set up the service level agreements (SLAs).
– A CSP inquires about the resource availability of other CSPs in the
federation, and requests a reservation of available resources that meet
the quality requirements of the CSC. The requested CSPs reply whether
or not they are able to lease the resources.
– The cloud resource management (e.g., CRUD: create, read, update and
delete) are operated across multiple CSPs. The management is to enable
the leasing of cloud resources from different CSPs in the federation.
– The relevant CSPs exchange monitoring and auditing information of the
leased resources.
High-level figure describing the use Pre-processes
• CSPs form a group with service level agreement (i.e., policy negotiation).
case • The P-CSP monitors CSC's quality of service.
1. CSCs accesses 3. If the P-CSP detects the service degradation
P-CSP services. and location change of the CSC, it chooses an
S-CSP nearer to the CSC, reserves the resource,
Primary- and performs migration to continue the service.
CSC CSP
2. The CSC moves. Y.3511(14)_FI.Tab3
o
S-CSP
CSC Secondary- S-CSP
CSP
Group
4. P-CSP changes the CSC
access to the S-CSP.
NOTE – After the actions, S-CSP monitors the CSC's quality of service.•
• P-CSP may take care of the CSC even after CSC moves.
• A single S-CSP is assumed as the recipient because the
migration is triggered by one CSC to be served by another CSP.
Derived requirements for the cloud The capability is required to support:
capability – Policy negotiation including SLA management among the multiple CSPs
within a pre-established group (i.e., federation);
– CSC service level monitoring at CSP. If the service level degrades, the CSP
should initiate the next configured actions;
– Discovery, reservation, use and release of cloud resources, based on CSC
location, in a dynamic manner (i.e., not relying on the pre-configuration)
on other CSPs within the federation;
– Capability migration (e.g., virtual machine (VM) and applications) over
the reserved resources on other CSPs within the federation;
– Alteration and reversion (i.e., switchover and switchback) of CSC access
to one CSP to another CSP in a dynamic manner (i.e., not relying on the
pre-configuration) within the federation;
– Exchange of monitoring and auditing information among the multiple
CSPs within the federation;
– Exchange of authentication information about CSC (user/enterprise)
authentication status among the multiple CSPs within the federation.
301