NOF-based read control method, apparatus, and system

1. A read control method comprising:

receiving, by a non-volatile memory express over fabric (NOF) engine, using a communication link, a read request sent by a host;

sending, by the NOF engine, at least one read command to a non-volatile memory express (NVMe) hard disk according to the read request;

generating, by the NOF engine in response to congestion occurring on the communication link, a congestion flag corresponding to the communication link; and

sending, by the NOF engine, the congestion flag to the NVMe hard disk, wherein the congestion flag instructs the NVMe hard disk to suspend processing of the read command corresponding to the communication link.

2. The method according to claim 1 , wherein the sending, by the NOF engine, the congestion flag to the NVMe hard disk comprises:

determining, by the NOF engine, a submission queue corresponding to the communication link, wherein the submission queue stores a plurality of Submission Queue Entry (SQE) commands in the submission queue corresponding to the communication link, and wherein a first SQE command comprises at least one of a read command or a write command; and

sending, by the NOF engine, in response to receiving a command read request that is of the NVMe hard disk and that is for the submission queue, a second SQE command carrying the congestion flag to the NVMe hard disk.

3. The method according to claim 2 , wherein the method further comprises performing, after the sending, to the NVMe hard disk, the second SQE command:

sending, by the NOF engine, an idle frame to the host at an interval using the communication link;

generating, by the NOF engine, in response to receiving an acknowledgment signal of the idle frame, a congestion release flag corresponding to the communication link; and

sending, by the NOF engine, the congestion release flag to the NVMe hard disk, wherein the congestion release flag instructs the NVMe hard disk to resume the processing of the read command corresponding to the communication link.

4. The method according to claim 3 , wherein the sending, by the NOF engine, the congestion release flag to the NVMe hard disk comprises:

determining, by the NOF engine, the submission queue corresponding to the communication link, wherein the submission queue stores the plurality of SQE commands corresponding to the communication link; and

sending, by the NOF engine, in response to receiving a command read request that is of the NVMe hard disk and that is for the submission queue, a third SQE command carrying the congestion release flag to the NVMe hard disk.

5. The method according to claim 3 , wherein the sending, by the NOF engine, the congestion release flag to the NVMe hard disk comprises:

determining, by the NOF engine, the submission queue corresponding to the communication link, wherein the submission queue stores the plurality of SQE commands corresponding to the communication link; and

sending, by the NOF engine, to the NVMe hard disk, a register write command pointing to a second user-defined address, wherein the second user-defined address is an address that stores the congestion release flag, and wherein the register write command carries an identifier of the submission queue and the congestion release flag.

6. The method according to claim 3 , wherein the sending, by the NOF engine, the congestion release flag to the NVMe hard disk comprises:

determining, by the NOF engine, the submission queue corresponding to the communication link, wherein the submission queue stores the plurality of SQE commands corresponding to the communication link; and

sending, by the NOF engine, a management command containing second user-defined content to the NVMe hard disk, wherein the second user-defined content carries an identifier of the submission queue and the congestion release flag.

7. The method according to claim 1 , wherein the sending, by the NOF engine, the congestion flag to the NVMe hard disk comprises:

determining, by the NOF engine, a submission queue corresponding to the communication link, wherein the submission queue stores a Submission Queue Entry (SQE) command corresponding to the communication link; and

sending, by the NOF engine, to the NVMe hard disk, a register write command pointing to a first user-defined address, wherein the first user-defined address is an address that stores the congestion flag, and wherein the register write command carries an identifier of the submission queue and the congestion flag.

8. The method according to claim 1 , wherein the sending, by the NOF engine, the congestion flag to the NVMe hard disk comprises:

determining, by the NOF engine, a submission queue corresponding to the communication link, wherein the submission queue stores a Submission Queue Entry (SQE) command corresponding to the communication link; and

sending, by the NOF engine, a management command containing first user-defined content to the NVMe hard disk, wherein the first user-defined content carries an identifier of the submission queue and the congestion flag.

9. A non-volatile memory express over fabric (NOF)-based read control method, comprising:

receiving, by a NOF engine having a data buffer corresponding to a communication link, using the communication link, a read request sent by a host, wherein the read request requests to read data of a first data amount, and wherein the first data amount exceeds a size of the data buffer corresponding to the communication link;

splitting, by the NOF engine, the read request into K read commands, wherein the read command requests to read data of a second data amount, and wherein the second data amount is less than the first data amount;

sending, by the NOF engine, m read commands to a non-volatile memory express (NVMe) hard disk, wherein m≤K, and a product of m and the second data amount is not greater than the size of the data buffer corresponding to the communication link; and

suspending, by the NOF engine in response to congestion occurring on the communication link, sending of a remaining read command to the NVMe hard disk.

10. The method according to claim 9 , wherein the method further comprises performing, after the suspending the sending of the remaining read command to the NVMe hard disk:

receiving, by the NOF engine, data read by the NVMe hard disk according to the m read commands;

storing, by the NOF engine, the data in the data buffer corresponding to the communication link; and

sending, by the NOF engine in response to the congestion on the communication link being released, the data in the data buffer to the host using the communication link.

11. The method according to claim 10 , wherein the method further comprises performing, after the sending the data in the data buffer to the host:

continuing, by the NOF engine, to send n read commands to the NVMe hard disk, wherein n≤K−m, and wherein a product of n and the second data amount is not greater than the size of the data buffer corresponding to the communication link.

12. A non-volatile memory express over fabric (NOF) engine, comprising:

at least one processor; and

a memory, storing a program to be executed by the at least one processor, the program including instructions for:

receiving, using a communication link, a read request sent by a host;

sending at least one read command to a non-volatile memory express (NVMe) hard disk according to the read request;

generating a congestion flag corresponding to the communication link in response to congestion occurring on the communication link; and

sending the congestion flag to the NVMe hard disk, wherein the congestion flag instructs the NVMe hard disk to suspend processing of the read command corresponding to the communication link.

13. The NOF engine according to claim 12 , wherein the instructions for sending the congestion flag to the NVMe hard disk include instructions for:

determining a submission queue corresponding to the communication link, wherein the submission queue stores a plurality of Submission Queue Entry (SQE) commands in the submission queue corresponding to the communication link, and wherein a first SQE command comprises at least one of a read command or a write command; and

sending, in response to receiving a command read request that is of the NVMe hard disk and that is for the submission queue, a second SQE command carrying the congestion flag to the NVMe hard disk.

14. The NOF engine according to claim 13 , wherein the program further includes instructions for performing, after the sending, to the NVMe hard disk, the second SQE command:

sending, by the NOF engine, an idle frame to the host at an interval using the communication link;

generating, by the NOF engine, in response to receiving an acknowledgment signal of the idle frame, a congestion release flag corresponding to the communication link; and

sending, by the NOF engine, the congestion release flag to the NVMe hard disk, wherein the congestion release flag instructs the NVMe hard disk to resume the processing of the read command corresponding to the communication link.

15. The NOF engine according to claim 14 , wherein the instructions for sending the congestion release flag to the NVMe hard disk include instructions for:

determining, by the NOF engine, the submission queue corresponding to the communication link, wherein the submission queue stores the plurality of SQE commands corresponding to the communication link; and

performing at least one of:

sending, in response to receiving a command read request that is of the NVMe hard disk and that is for the submission queue, a third SQE command carrying the congestion release flag to the NVMe hard disk;

sending, to the NVMe hard disk, a register write command pointing to a second user-defined address, wherein the second user-defined address is an address that stores the congestion release flag, and wherein the register write command carries an identifier of the submission queue and the congestion release flag; or

sending, a management command containing second user-defined content to the NVMe hard disk, wherein the second user-defined content carries the identifier of the submission queue and the congestion release flag.

16. The NOF engine according to claim 12 , wherein the instructions for sending the congestion flag to the NVMe hard disk include instructions for:

determining a submission queue corresponding to the communication link, wherein the submission queue stores a Submission Queue Entry (SQE) command corresponding to the communication link; and

sending, by the NOF engine, to the NVMe hard disk, a register write command pointing to a first user-defined address, wherein the first user-defined address is an address that stores the congestion flag, and wherein the register write command carries an identifier of the submission queue and the congestion flag.

17. The NOF engine according to claim 12 , wherein the instructions for sending, by the NOF engine, the congestion flag to the NVMe hard disk include instructions for:

determining, by the NOF engine, a submission queue corresponding to the communication link, wherein stores a Submission Queue Entry (SQE) command corresponding to the communication link; and

sending, by the NOF engine, a management command containing first user-defined content to the NVMe hard disk, wherein the first user-defined content carries an identifier of the submission queue and the congestion flag.

18. A non-volatile memory express over fabric (NOF) engine, comprising:

at least one processor; and

a memory, wherein the memory comprises a data buffer corresponding to a communication link, wherein the memory stores a program to be executed by the at least one processor, the program including instructions for:

receiving, using the communication link, a read request sent by a host, wherein the read request requests to read data of a first data amount, and wherein the first data amount exceeds a size of the data buffer of the communication link;

splitting the read request into K read commands, wherein the read command requests to read data of a second data amount, and wherein the second data amount is less than the first data amount;

sending m read commands to a non-volatile memory express (NVMe) hard disk, wherein m≤K, and wherein a product of m and the second data amount is not greater than the size of the data buffer of the communication link; and

suspending sending of a remaining read command to the NVMe hard disk in response to congestion occurring on the communication link.

19. The NOF engine according to claim 18 , wherein the program further includes instructions for performing, after the suspending the sending of the remaining read command to the NVMe hard disk:

receiving data read by the NVMe hard disk according to the m read commands;

storing the data in the data buffer corresponding to the communication link; and

sending, in response to the congestion on the communication link being released, the data in the data buffer to the host using the communication link.

20. The NOF engine according to claim 19 , wherein the program further includes performing, after the sending the data in the data buffer to the host:

continuing, by the NOF engine, to send n read commands to the NVMe hard disk, wherein n≤K−m, and wherein a product of n and the second data amount is not greater than the size of the data buffer corresponding to the communication link.