Nested Suspend Resume Mechanism for SSD Read Latency

The patent segments write and erase commands into multiple smaller sub-commands or chunks. Instead of executing a single large write or erase command that blocks reads, the command is divided into smaller units that can be interleaved with read operations. This segmentation allows the SSD to process reads between sub-command executions, reducing read latency while maintaining overall productivity through continuous segmented operation.

The patent implements dynamic command scheduling that adjusts the number and type of outstanding commands based on current workload conditions and QoS requirements. The system dynamically monitors command completion times and adjusts the command queue depth, suspending or prioritizing commands to maintain consistent completion times. This dynamic adaptation allows the system to achieve high throughput when conditions permit while ensuring QoS compliance when latency consistency is critical.

The patent applies different quality levels of service to different command types based on their requirements. Read commands receive high priority with guaranteed responsiveness and low latency, while write and erase commands are executed in the background with lower priority. The system locally optimizes resource allocation for each command type, ensuring reads get the necessary bandwidth and responsiveness while writes and erases maintain acceptable throughput through non-blocking execution.

The patent implements a nested suspend/resume mechanism where suspension states are organized in hierarchical levels. When a read command arrives, the system suspends lower-priority write or erase commands, and if those commands have nested sub-commands, those are suspended in turn. This nested structure allows the system to systematically pause operations at multiple levels without requiring complex global coordination, reducing control logic complexity while maintaining QoS consistency through structured priority management.