Cloud Service Deployment with Resource Reservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In cloud computing systems, service deployment failures often occur due to the unavailability of computing resources, leading to extensive rollbacks and delays in resource allocations, negatively impacting deployment performance and customer experience.
Innovation Solution
A method that reserves computing resources first with an expiration provision, allowing for parallel determination of resource availability and independent instantiation processing, thereby dissociating scarce computing resource requirements from other resource requirements and reducing workflow failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If computing resources are allocated first followed by other resources (networking, storage), then resource allocation can proceed in sequence, but deployment failures occur when resources are unavailable causing rollbacks and delays
Solution Approach 1:
The patent applies preliminary action by reserving computing resources before initiating the full deployment workflow. The system performs a preliminary check to determine if computing resources are available, and if so, reserves them for a predetermined period. This allows the deployment to proceed with confidence that critical resources are secured, reducing the likelihood of failure and subsequent rollback. The reservation acts as a preparatory step that ensures resource availability before committing to the full deployment process.
2Reliability
If all resources are committed/allocated before service activation, then resource availability is ensured, but failure of any resource requires rollback of all committed resources
Solution Approach 1:
The patent segments the resource allocation process into distinct phases: computing resource reservation, other resource location, and service activation. Instead of treating all resources as a single committed unit, the system separates computing resources (which are scarce and critical) from other resources (networking, storage). This segmentation allows the deployment to proceed independently for computing resources while other resources are being located, and enables partial success scenarios where computing resources are secured even if other resources take longer to provision.
Solution Approach 2:
The system performs preliminary reservation of computing resources before completing the allocation of all resources. This preliminary action secures the critical path resources (computing) while allowing non-critical resources (networking, storage) to be located and allocated in parallel or subsequently. The reservation expires after a predetermined period if deployment is not completed, automatically releasing resources if the process stalls.
3Quantity of substance
If computing resources are scarce compared to networking and storage resources, then computing resource availability becomes the bottleneck, but traditional allocation methods do not prioritize scarce resources
Solution Approach 1:
The patent applies local quality by treating different resource types differently based on their availability characteristics. Computing resources, being scarce and critical, receive special treatment through reservation mechanisms with predetermined expiration periods. Other resources (networking, storage) which are more abundant are handled through standard location and allocation processes. This differentiated approach ensures that the bottleneck resource (computing) is prioritized and secured first, while other resources are allocated in parallel or subsequently without blocking the critical path.
Data Source
AI summary
A system receives a request for a plurality of resources of a cloud computing system to deploy a service. The system selects a portion of the resources and determines whether the selected portion is at least partially available in one or more clusters in parallel. If available, the system reserves the available resources for a predetermined period of time. If unavailable in one of the clusters, the system releases the reserved resources after the predetermined period of time. While or after determining the availability, the system locates at least a portion of the remainder of the resources in the one or more clusters. The system allocates the reserved resources on one of the clusters and the located remainder of the resources to deploy the service and instantiates the service on the one of the clusters using the allocated resources.