Thin-Provisioned Cache Volumes for Fault-Domain Write Resiliency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Disaggregated storage systems face temporary unavailability, leading to degraded data resiliency and increased resource consumption to maintain data availability, as write operations may fail to reach all data copies.
Innovation Solution
A computer system dynamically maintains storage resiliency using thinly provisioned cache volumes, allocating additional storage on-demand at available fault domains when some domains become unavailable, and caching extra data copies to ensure redundancy without permanently increasing hardware usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If additional storage resources are obtained to store redundant copies of data, then data availability and resiliency are improved, but resource consumption increases
Solution Approach 1:
The patent implements dynamic cache allocation where cache volumes are provisioned temporarily only when fault domains become unavailable. During healthy operation, no cache storage is allocated. When unavailability is detected, cache volumes are dynamically created to store redundant data copies. This dynamic approach allows the system to maintain high data availability during failures without permanently consuming additional storage resources, directly resolving the contradiction between reliability improvement and resource consumption.
2Reliability
If cache volumes are permanently allocated to maintain data copies, then data resiliency is improved, but storage footprint increases during healthy operation
Solution Approach 1:
The system employs periodic health checks of fault domains and dynamically activates cache volumes only during periods when unavailability is detected. Cache volumes are created temporarily to maintain data resiliency during these periods, then deallocated when health is restored. This periodic activation pattern ensures data resiliency is maintained when needed while minimizing storage footprint during healthy operation, directly addressing the contradiction between resiliency improvement and footprint increase.
3Productivity
If redundant storage capacity is pre-provisioned, then write operations can maintain multiple data copies, but resource efficiency decreases
Solution Approach 1:
The system performs preliminary health checks of fault domains before executing write operations. When unavailability is detected in advance, the system proactively creates cache volumes and redirects write operations to store data copies in available fault domains. This preliminary detection and proactive cache creation ensures write operations can successfully maintain multiple data copies without requiring pre-provisioned redundant storage capacity, thereby improving write operation success rate while maintaining resource efficiency.
Data Source
AI summary
A computer system is configured to provision a plurality of storage volumes at a plurality of fault domains and thinly provision a plurality of cache volumes at the plurality of fault domains. The computer system is also configured to perform a write operation in a resilient manner that maintains a plurality of copies of data associated with the write operation. Performing the write operation in the resilient manner includes allocating a portion of storage in each of the plurality of cache volumes, and caching the data associated with the write operation in the portion of storage in each of the plurality of cache volumes. The cached data is then persistently stored in the plurality of storage volumes. After that, the portion of storage in each of the plurality of cache volumes is deallocated.