System and method for managing tasks in a storage device
A method for scheduling requests in a storage system includes receiving, at a scheduler, a request associated with a first storage device, receiving, at the scheduler, task information from a neural network circuit, the task information being associated with a task performed by the first storage device, and modifying a processing of the request at the first storage device based on the task information.
1. A method of scheduling requests in a storage system, the method comprising:
receiving, at a scheduler, a request associated with a first storage device;
receiving, at the scheduler, task information from a neural network circuit, the task information indicating a likelihood of the first storage device performing a task; and
modifying a processing of the request at the first storage device based on the task information, such that an operation associated with the request is performed at a physical location that is external to the first storage device.
2. The method of claim 1 , wherein:
the request is an input/output (I/O) request;
the scheduler is an I/O scheduler;
the task is a background task; and
the task information comprises an indication that the first storage device has a probability of performing the background task that exceeds a threshold.
3. The method of claim 1 , wherein the task comprises a garbage collection task or a wear levelling task.
4. The method of claim 1 , wherein:
the request comprises a read request; and
the modifying the processing of the request at the first storage device comprises:
prefetching data from the first storage device based on the task information;
storing the data at a buffer associated with the scheduler; and
retrieving at least a portion of the data from the buffer based on the request and based on the task information.
5. The method of claim 4 , further comprising:
determining, by the scheduler, a status of the task based on information received via a firmware interface of the first storage device; and
moving the data from the buffer to the first storage device based on the status.
6. The method of claim 1 , wherein:
the request comprises a write request; and
the modifying the processing of the request at the first storage device comprises rerouting the processing of the request from the first storage device to a second storage device based on the task information.
7. The method of claim 1 , wherein the neural network circuit is trained to generate the task information based on:
a simulation circuit configured to trace operations from applications running on a host and to generate a trace based on the operations; and
a machine learning (ML) model trainer configured to input data associated with the trace to the neural network circuit.
8. A system for scheduling requests, the system comprising:
a processor; and
a memory storing instructions, which, based on being executed by the processor, cause the processor to perform:
receiving a request associated with a first storage device;
receiving task information from a neural network circuit, the task information indicating a likelihood of the first storage device performing a task; and
modifying a processing of the request at the first storage device based on the task information, such that an operation associated with the request is performed at a physical location that is external to the first storage device.
9. The system of claim 8 , wherein:
the request is an input/output (I/O)) request;
the task is a background task; and
the task information is generated based on a workload associated with the first storage device.
10. The system of claim 8 , wherein the task comprises a garbage collection task or a wear levelling task.
11. The system of claim 8 , wherein:
the request comprises a read request; and
the modifying the processing of the request at the first storage device comprises:
prefetching data from the first storage device based on the task information;
storing the data at a buffer; and
retrieving at least a portion of the data from the buffer based on the request and based on the task information.
12. The system of claim 11 , wherein the instructions, based on being executed by the processor, further cause the processor to perform:
determining a status of the task based on information received via a firmware interface of the first storage device; and
moving the data from the buffer to the first storage device based on the status.
13. The system of claim 8 , wherein:
the request comprises a write request; and
the modifying the processing of the request at the first storage device comprises rerouting the processing of the request from the first storage device to a second storage device based on the task information.
14. The system of claim 8 , wherein the neural network circuit is trained to generate the task information based on:
a simulation circuit configured to trace operations from applications running on a host and generate a trace based on the operations; and
a machine learning (ML) model trainer configured to input data associated with the trace to the neural network circuit.
15. A storage system comprising:
a neural network circuit; and
a scheduler configured to be communicably coupled to the neural network circuit and a first storage device, the scheduler being further configured to:
receive a request associated with the first storage device;
receive task information from the neural network circuit, the task information indicating a likelihood of the first storage device performing a task; and
modify a processing of the request at the first storage device based on the task information, such that an operation associated with the request is performed at a physical location that is external to the first storage device.
16. The storage system of claim 15 , wherein:
the request is an I/O request;
the scheduler is an I/O scheduler;
the task is a background task; and
the storage system comprises an all-flash array.
17. The storage system of claim 15 , wherein the task comprises a garbage collection task or a wear levelling task.
18. The storage system of claim 15 , wherein:
the request comprises a read request; and
the modifying the processing of the request at the first storage device comprises:
prefetching data from the first storage device based on the task information;
storing the data at a buffer associated with the scheduler; and
retrieving at least a portion of the data from the buffer based on the request and based on the task information.
19. The storage system of claim 18 , wherein the scheduler is configured to:
determine a status of the task based on information received via a firmware interface of the first storage device; and
move the data from the buffer to the first storage device based on the status.
20. The storage system of claim 15 , wherein:
the request comprises a write request; and
the modifying the processing of the request at the first storage device comprises rerouting the processing of the request from the first storage device to a second storage device based on the task information.