Arbitration management method of task queue, storage medium and solid state disk

Abstract

The invention relates to the technical field of data storage, and provides a task queue arbitration management method, a storage medium and a solid state disk, the method comprises the following steps: judging whether a first task queue with an emergency queue priority is empty or not; if the task queue with the emergency queue priority is not empty, the first task queue is scheduled preferentially; in the process of scheduling the first task queue, monitoring whether the scheduling state of the task queue meets a preset priority scheduling right release condition or not; and when the scheduling state of the task queue meets a preset priority scheduling right release condition, scheduling the task queues of which the queue priorities are other levels according to a weighted cycle scheduling strategy. According to the invention, the WRR queue can be prevented from being starved during scheduling, and meanwhile, the priority scheduling of the emergency priority can be provided for the upper layer service, so that the emergency priority queue can be widely used in the actual service, and the flexibility of adjusting the priority by the service is greatly improved.

Description

technical field [0001] The invention relates to the technical field of data storage, in particular to an arbitration management method of a task queue, a storage medium and a solid-state hard disk. Background technique [0002] NVMe SSD uses multi-queue technology to complete efficient IO requests. There are generally two methods for multi-queue scheduling and arbitration in SSD controllers: RR (Round Robin, round-robin scheduling) and Weighted Round Robin (WRR, weighted round-robin scheduling). As upper-layer services have more and more priority requirements for specific IO services, the RR arbitration mechanism can no longer meet the needs of services with priority requirements. Therefore, the weighted WRR and urgent class emergency queue arbitration mechanisms are becoming more and more popular. is widely used. However, the WRR and urgent technologies mentioned in the NVMe protocol have certain defects and cannot fully meet the needs of the market. [0003] The specific...

AI Technical Summary

Problems solved by technology

[0005] 2. Urgent class queues have the highest priority, but due to the problem of starving other queues during use, they are rarely used in actual business at present

[0006] 3. According to the description of the NVMe spec, in the existing technical solutions, the queue priority cannot be changed after it is created. If you need to change it, you can only delete the queue and recreate the queue

The existing technical solutions cannot adjust the weight ratio of the urgent queue to obtain different service effects, and when the number of requested services is too large, it may also cause the problem of starvation

[0010] 3. The priority queue is specified when creating the SQ (submission queue). If you want to re-adjust the priority, you must delete the original SQ and then create it, that is, the upper-level business must stop

For the upper-level business that needs to adjust the priority at runtime, the way to delete the queue and create it is not flexible, and it will greatly reduce the availability

Images