Multicast Address Groups for Persistent Memory Data Writes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional data storage systems face increased latency due to the need for multiple software, firmware, and hardware processing layers, and creating multiple copies for fault resiliency further complicates data transfer, leading to inefficiencies in writing and reading data from persistent storage.
Innovation Solution
A method and storage appliance that utilize pre-defined multicast address groups to dynamically select appropriate storage locations for writing data, ensuring efficient storage of multiple copies with low latency by using an IO fabric configured with multicast addresses, taking into account the current state of storage devices, thereby adapting to changes in storage configurations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple copies of data are created by the storage system to provide increased fault resiliency, then reliability is improved, but latency increases beyond the latency required to write to the underlying storage medium once
Solution Approach 1:
The system pre-establishes multicast address groups and maps them to multiple storage devices before data write operations. When a write operation occurs, the system uses the pre-configured multicast mechanism to simultaneously write data to multiple storage devices in parallel, eliminating sequential write operations and reducing latency while maintaining fault resiliency through replicated data copies.
Solution Approach 2:
The patent combines multiple data write operations into a single multicast write operation. Instead of writing data to multiple storage devices sequentially through multiple separate write calls, the system uses a single write operation with a multicast address group that simultaneously delivers data to all member storage devices, thereby reducing latency while achieving the same reliability goal.
2Productivity
If traditional approaches are used to transfer data to persistent storage, then data can be written to storage, but multiple layers of software, firmware, and hardware processing are required which increases complexity and latency
Solution Approach 1:
The patent extracts and eliminates unnecessary intermediate processing layers by implementing a direct write path from the host to storage devices through the multicast mechanism. By using the existing IO fabric and address translation mechanisms already present in the system, the invention avoids adding new software or firmware layers that would increase complexity, while still achieving efficient data transfer to multiple storage devices.
3Reliability
If the storage system creates multiple copies of written data to provide fault resiliency, then data reliability is improved, but extra latency is induced beyond the latency required to write to the underlying storage medium once
Solution Approach 1:
The system pre-establishes multicast address groups and maps them to multiple storage devices before data write operations. When a write operation occurs, the system uses the pre-configured multicast mechanism to simultaneously write data to multiple storage devices in parallel, eliminating sequential write operations and reducing latency while maintaining fault resiliency through replicated data copies.
Solution Approach 2:
The patent combines multiple data write operations into a single multicast write operation. Instead of writing data to multiple storage devices sequentially through multiple separate write calls, the system uses a single write operation with a multicast address group that simultaneously delivers data to all member storage devices, thereby reducing latency while achieving the same reliability goal.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
In general, the technology relates to a method and system for writing data to persistent storage. More specifically, embodiments of the technology relate to writing data to vaulted memory segments in persistent storage using pre-defined multicast address groups. Further, embodiments of the technology take into account the current state of the persistent storage in order to select the vaulted memory segments in which to store the data.