Updating No Sync Technique for Continuous Storage Service
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In a distributed storage cluster, temporary storage capacity limitations hinder continuous data protection and availability when one or more storage nodes fail, as incoming data cannot be efficiently written due to exhausted cache resources.
Innovation Solution
The Updating No Sync (UNS) technique coopts healthy block services to temporarily store data updates during a degraded cluster state, allowing overflow data to be accumulated without synchronization, and synchronizes this data back once the cluster is restored, thereby extending storage service continuity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If temporary storage is used to store incoming data during node failure, then data protection is improved, but storage capacity is limited and becomes exhausted
Solution Approach 1:
The patent segments the storage system into multiple storage nodes organized in a cluster, where data is distributed across multiple nodes rather than relying on a single temporary storage location. This segmentation allows the system to leverage the combined storage capacity of all healthy nodes, effectively overcoming the limitation of individual temporary storage capacity while maintaining data protection through distributed redundancy.
2Duration of action of moving object
If temporary storage capacity is increased to handle overflow data, then storage service continuity is improved, but device complexity and cost increase
Solution Approach 1:
The patent makes healthy storage nodes serve multiple functions: they act as both data storage locations and temporary overflow destinations during failures. Instead of requiring dedicated cache infrastructure, existing storage nodes are co-opted to handle overflow data, eliminating the need for separate large-capacity cache devices while maintaining service continuity through the multi-functional use of cluster resources.
3Reliability
If data is written to temporary storage during degraded state, then data availability is improved, but synchronization complexity increases when cluster is restored
Solution Approach 1:
The patent introduces a bin assignment service as an intermediary that manages the coordination between slice services and block services during degraded and restored states. This intermediary handles the complex synchronization logic by tracking which nodes are healthy, determining appropriate data placement, and managing the restoration process, thereby simplifying the overall system architecture while maintaining data availability and enabling automated synchronization.
Data Source
AI summary
An Updating No Sync (UNS) technique ensures continuous data protection for content driven distribution of data served by storage nodes of a fault tolerant cluster in the event of a degraded cluster state. A storage service implemented in each node includes one or more slice services (SSs) configured to process and store metadata describing the data served by the storage nodes and one or more block services (BSs) configured to process and store the data on storage devices of the node. A bin assignment service may coopt one or more healthy BSs to temporarily store updates of data and metadata received at the SS while the cluster is in the degraded state as an overflow data path (hence the term “Updating No Sync,” which denotes updating without synchronizing, i.e., not distributing the data within the cluster, but only accumulating an overflow of SS information). Once the cluster is no longer degraded, the accumulated overflow SS information at the BSs may be synchronized back to restored BSs, i.e., according to write path determination in the absence of node failure/unavailability.