Distributed Memory Redundancy via Fabric Link Failover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large-scale distributed memory systems face reliability issues due to increased failure points as the number of components grows, leading to media failures, power delivery issues, and other faults that can cause server reboots and reduce availability.
Innovation Solution
The implementation of redundant memory devices and fabrics in distributed memory systems, where two memory devices are paired to store the same data, with write requests duplicated and read requests serviced from either device, and the use of a memory processing unit (MPU) to manage these paired devices and alternative fabric links for failure tolerance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the number of components in distributed memory systems increases to expand system capacity, then the system capacity and functionality are improved, but the number of failure points increases leading to reduced reliability
Solution Approach 1:
The system is divided into multiple independent fabric links connecting memory devices in pairs. Each fabric link operates independently, and failure of one link does not affect others. The memory devices are segmented into primary and secondary roles, with dedicated fabric links for each, isolating failure domains and maintaining system reliability while expanding capacity.
Solution Approach 2:
Redundant memory devices are pre-configured in paired arrangements with dedicated fabric links before failures occur. This proactive redundancy ensures that when a component fails, the system has already prepared alternative paths and devices to maintain operation, cushioning against the reliability degradation that would otherwise result from increased component count.
2Reliability
If redundant memory devices are implemented to improve failure tolerance, then system reliability is improved, but device complexity increases
Solution Approach 1:
Multiple fabric links are merged into a unified fabric manager that handles routing and failover logic centrally. The memory device pairs are merged at the logical level through a single namespace, presenting a simplified interface to hosts while maintaining physical redundancy. This consolidation reduces the apparent complexity despite the underlying redundant hardware.
Solution Approach 2:
The fabric manager serves multiple functions: it manages normal data routing, monitors fabric link health, detects failures, and executes failover operations. This multi-functionality consolidates what would otherwise require separate dedicated components for each function, reducing overall device complexity while maintaining comprehensive failure tolerance.
3Reliability
If fabric links are increased to provide alternative paths for failure tolerance, then system availability is improved, but device complexity and power consumption increase
Solution Approach 1:
The fabric link configuration is dynamic rather than static. Under normal conditions, only the primary fabric link is actively used for data transmission, consuming minimal power. When failures are detected, the system dynamically activates alternative fabric links. This dynamic activation ensures high availability while minimizing power consumption during normal operation.
Solution Approach 2:
The primary fabric link maintains continuous active operation for normal data transmission, while alternative links remain in a low-power standby state. This ensures that the useful action of data transmission continues uninterrupted through the primary path, while backup paths consume minimal power until needed, balancing availability with energy efficiency.
Data Source
AI summary
Disclosed in some examples are methods, systems, devices, and architectures which provide for techniques for memory device and memory fabric redundancy within distributed memory systems. In some examples, two memory devices are paired and each stores a same set of data such that writes to the memory devices are duplicated and reads may be satisfied from either device. In some examples, a memory processing unit (MPU) may be incorporated into the memory architecture to support these paired memory devices. The MPU may be placed between the host and a multi-planed memory fabric which connects to multi-ported CXL memory devices. In some examples, the MPU may also enable the use of alternative fabric links. That is, if a memory fabric link between the MPU and a memory device is unavailable, an alternative link may be utilized to restore connectivity to a memory device.