Systems and method for batching requests in computational devices
A system is described. The system may include a processor. Software may run on the processor to build a batch request, which may include at least a first request and a second request. The system may also include a storage device and a computational storage unit associated with the storage device, which may be configured to execute the batch request.
1. A system, comprising:
a chassis;
a processor in the chassis;
software configured to run on the processor and to build a batch request for an application, the batch request including at least a first request and a second request;
a storage device in the chassis; and
a computational storage unit in the chassis associated with the storage device, the computational storage unit configured to process the batch request using data stored on the storage device received from the application.
2. The system according to claim 1 , wherein the computational storage unit is configured to receive the batch request from the application running on the processor and to return a batch result of the batch request to the application based at least in part on the first request and the second request.
3. The system according to claim 1 , wherein the software includes request add software to add the first request to the batch request.
4. The system according to claim 1 , wherein the batch request includes one of a serial structure, a parallel structure, or a hybrid structure.
5. The system according to claim 1 , wherein:
the software includes a first software; and
the system further comprises second software configured to send the batch request to the computational storage unit.
6. The system according to claim 1 , wherein the first request and the second request are included in the batch request based at least in part on the first request and the second request being related.
7. The system according to claim 1 , wherein the computational storage unit includes:
a receiver to receive the batch request from the application running on the processor;
a parser to parse the batch request into at least the first request and the second request;
a graph constructor to construct a directed graph from at least the first request and the second request;
a scheduler to schedule the first request and the second request for execution based at least in part on the directed graph; and
hardware to execute at least the first request.
8. A computational storage unit, comprising:
a receiver to receive a batch request from an application running on a processor in a chassis;
a parser to parse the batch request into at least a first request and a second request;
a graph constructor to construct a directed graph from at least the first request and the second request;
a scheduler to schedule the first request and the second request for execution based at least in part on the directed graph;
hardware to execute the first request based at least in part on the scheduler; and
a transmitter to send a batch result of the batch request to the application running on the processor based at least in part on the first request and the second request,
wherein the computational storage unit is in the chassis.
9. The computational storage unit according to claim 8 , further comprising second hardware to execute the second request.
10. The computational storage unit according to claim 8 , further comprising a verifier to verify that the directed graph is acyclic.
11. The computational storage unit according to claim 8 , wherein the scheduler is configured to schedule the first request to execute based at least in part on the directed graph indicating that the first request is independent of the second request.
12. The computational storage unit according to claim 8 , wherein the scheduler is configured to schedule the first request to execute based at least in part on a completion of the second request and the directed graph indicating that the first request depends on the second request.
13. A method, comprising:
receiving, at a software, a batch create request from an application to create a batch request;
receiving, at the software, a first add request from the application to add a first request to the batch request;
receiving, at the software, a second add request from the application to add a second request to the batch request; and
storing, by the software, the batch request,
wherein the software is separate from the application.
14. The method according to claim 13 , wherein receiving, at the software, the second add request from the application to add the second request to the batch request includes receiving, at the software, the second add request from the application to add the second request to the batch request in series with the first request.
15. The method according to claim 13 , wherein receiving, at the software, the second add request from the application to add the second request to the batch request includes receiving, at the software, the second add request from the application to add the second request to the batch request in parallel with the first request.
16. The method according to claim 13 , wherein receiving, at the software, the first add request from the application to add the first request to the batch request includes:
receiving, at the software, from the application, an identifier of an input buffer for the first request; and
receiving, at the software, from the application, an identifier of an output buffer for the first request.
17. The method according to claim 13 , further comprising configuring, by the software, the batch request.
18. The method according to claim 17 , wherein configuring, by the software, the batch request includes setting, by the software, a parameter of the batch request to a value.
19. The method according to claim 13 , wherein the batch request is independent of a hardware of a storage device or a computational storage unit associated with the storage device.
20. The method according to claim 13 , wherein a computational storage unit is configured to execute the batch request on behalf of the application on a first data and to execute the batch request on behalf of a second application on a second data.
21. The method according to claim 13 , wherein:
the first request is to process a first data into a second data; and
the second request is to process the second data into a third data.