Memory Load Balancing via Dynamic Internal Request Throttling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional storage systems face performance degradation due to the significant consumption of I/O bandwidth by internally-generated memory operations, leading to increased latency and poor user experience, especially in hybrid systems combining Hard Disk Drives (HDDs) and Solid State Drives (SSDs).
Innovation Solution
Implementing a method to balance internally-generated I/O operations with externally-generated I/O operations by regulating the rate of internal memory requests based on external request loads, using batch processing and load threshold determination to optimize HDD and SSD usage, thereby preventing bandwidth overload and maintaining performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If internally-generated I/O operations are allowed to run at full capacity, then storage system functionality is maintained, but external I/O performance degrades significantly
Solution Approach 1:
The system dynamically adjusts the rate at which internal memory operation requests are transferred to non-volatile memory based on real-time external I/O load conditions. When external I/O demand is high, internal request transfer rate is reduced; when external demand is low, internal transfer rate increases. This dynamic rate regulation resolves the contradiction by making the internal I/O throughput adaptive rather than static.
Solution Approach 2:
The system periodically evaluates external I/O load conditions and adjusts internal request processing accordingly. By continuously monitoring external I/O patterns and periodically regulating internal request rates, the system maintains storage functionality while preventing sustained degradation of external performance.
2Productivity
If internal memory requests are processed immediately, then storage operations are completed efficiently, but external request latency increases
Solution Approach 1:
The system dynamically controls the throughput of internal memory requests based on external load conditions. Instead of processing all internal requests at maximum speed, the system adjusts the transfer rate to match external demand, thereby reducing queueing delays for external requests while still maintaining adequate progress on internal storage operations.
3Productivity
If HDD head movement is optimized for internal requests, then internal I/O performance improves, but external random I/O performance degrades
Solution Approach 1:
The system dynamically regulates the rate of internal request transfer to HDD based on external I/O conditions. When external random I/O demand is high, the internal request rate is throttled to prevent head movement optimization from degrading external performance. This dynamic rate control allows the system to switch between optimizing for internal throughput and external response time.
Data Source
AI summary
The various embodiments described herein include methods, devices, and systems for processing memory operation requests. In one aspect, a method is performed at a computing system having one or more processors and non-volatile memory: (1) obtaining a plurality of internal memory operation requests for the non-volatile memory, the plurality of internal memory operation requests originating from within the computing system; (2) obtaining a plurality of external memory operation requests for the non-volatile memory, the plurality of external memory operation requests originating from one or more devices remote and distinct from the computing system; and (3) regulating a rate at which the plurality of internal memory operation requests are transferred to the non-volatile memory based on an amount of external memory operation requests in the plurality of external memory requests.