Partially Synchronized Writes for Low-Latency Redundant Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In data centers, the process of writing data to multiple storage devices for redundancy often results in increased latency due to the need for multiple acknowledgments, which can negatively impact the quality of service for workloads.
Innovation Solution
Implementing partially synchronized writes, where a compute node receives an initial acknowledgment and resumes execution before waiting for subsequent acknowledgments from other nodes, allowing data to be written to power loss protected buffers before being committed to non-volatile storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the compute node waits for confirmations from multiple networked data storage devices before resuming execution, then data durability is improved, but workload latency increases
Solution Approach 1:
The patent applies partial action by requiring only a threshold number of acknowledgments (e.g., 2 out of 3) rather than waiting for all storage devices to confirm. This allows the compute node to resume workload execution after receiving sufficient confirmations to ensure data durability, without waiting for all devices, thereby reducing latency while maintaining reliability
Solution Approach 2:
The patent implements preliminary action by sending write requests to multiple data storage devices simultaneously rather than sequentially. This parallel approach allows the compute node to receive acknowledgments from multiple devices concurrently, reducing the total time waiting for confirmations while ensuring data is replicated across multiple devices for durability
2Reliability
If the compute node sends data blocks to multiple data storage devices through the network, then data redundancy is improved, but quality of service deteriorates
Solution Approach 1:
The patent segments the write operation into independent parallel tasks by sending data blocks to multiple data storage devices simultaneously through the network. Each storage device processes the write independently, allowing the system to achieve redundancy without serializing the operations, thus maintaining quality of service while improving reliability
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Technologies for managing partially synchronized writes include a managed node. The managed node is to issue a write request to write a data block, on behalf of a workload, to multiple data storage devices connected to a network, pause execution of the workload, receive an initial acknowledgment associated with one of the multiple data storage devices, wherein the initial acknowledgement is indicative of successful storage of the data block, and resume execution of the workload after receipt of the initial acknowledgement and before receipt of subsequent acknowledgements associated with any of the other data storage devices. Other embodiments are also described and claimed.