Method and system for fast ordered writes with target collaboration
In general, embodiments of the technology relate to a method and system for performing fast ordered writes in a storage appliance that includes multiple separate storage modules. More specifically, embodiments of the technology enable multicasting of data to multiple storage modules in a storage appliance, where the order in which the write requests are processed is the same across all storage modules in the storage appliance.
1. A method for processing write requests comprising:
receiving, by a first storage module, a first commit request from a client, wherein the first commit request is associated with a first data and comprises an address, wherein the address is generated by applying a consistent hashing function to at least an object identifier and an offset associated with the first data, wherein the address is associated with the first storage module;
in response to receiving the first commit request:
processing the first commit request, wherein processing the first commit request causes a first copy of the first data to be stored in persistent storage in the first storage module, wherein the first copy of the first data was received from the client and by the first storage module prior to the first storage module receiving the first commit request;
generating, by the first storage module, a second commit request for a second storage module;
transmitting the second commit request to the second storage module;
receiving a first commit reply from the second storage module indicating that a second copy of the first data is stored in persistent storage in the second storage module, wherein the second copy of the first data was received from the client and by the second storage module prior to the first storage module receiving the first commit request;
in response to receiving the first commit reply and after processing the first commit request, generating, by the first storage module, a second commit reply corresponding to the first commit request; and
transmitting the second commit reply to the client.
2. The method of claim 1 , further comprising:
receiving a third commit request by the first storage module from the second storage module, wherein the third commit request is associated with a second data; and
processing the third commit request, wherein processing the third commit request causes a first copy of the second data to be stored in a slave partition of the persistent storage in the first storage module,
wherein the slave partition is associated with a master partition on the second storage module.
3. The method of claim 2 , further comprising:
prior to receiving the second commit request:
receiving a second copy of the second data from the client; and
temporarily storing the second copy of the second data in a buffer in the first storage module.
4. The method of claim 1 , wherein processing the first commit request comprises:
storing the first commit request in a global submission queue, wherein the global submission queue maintains an order of all commit requests received by the first storage module,
wherein the first commit request is processed based on the order.
5. The method of claim 1 , wherein processing the first commit request comprises:
storing the first commit request in a client submission queue associated with the client, wherein the first commit request is associated with a global sequencing tag,
wherein the first commit request is processed based on an order specified by the global sequencing tag.
6. The method of claim 1 , wherein the first commit request is received from the client from a fabric operatively connected to the client and the first storage module.
7. The method of claim 6 , wherein the second commit request is transmitted to the second storage module using the fabric.
8. The method of claim 6 , wherein the fabric is a Peripheral Component Interconnect Express (PCIe) fabric.
9. The method of claim 1 , wherein the persistent storage in the first storage module comprises phase change memory.
10. A storage appliance, comprising:
a fabric;
a storage array operatively connected to the fabric and comprising a first storage module and a second storage module,
wherein the first storage module is configured to:
receive, via the fabric, a first commit request from a client, wherein the first commit request is associated with a first data and comprises an address, wherein the address is generated by applying a consistent hashing function to at least an object identifier and an offset associated with the first data, wherein the address is associated with the first storage module;
in response to receiving the first commit request:
process the first commit request, wherein processing the first commit request causes a first copy of the first data to be stored in persistent storage in the first storage module, wherein the first copy of the first data was received from the client and by the first storage module prior to the first storage module receiving the first commit request;
generate a second commit request for a second storage module;
transmit, via the fabric, the second commit request to the second storage module;
receive, via the fabric, a first commit reply from the second storage module indicating that a second copy of the first data is stored in persistent storage in the second storage module, wherein the second copy of the first data was received from the client and by the second storage module prior to the first storage module receiving the first commit request;
in response to receiving the first commit reply and after processing the first commit request, generate a second commit reply corresponding to the first commit request; and
transmit, via the fabric, the second commit reply to the client.
11. The storage appliance of claim 10 , wherein the first storage module is further configured to:
receive a third commit request from the second storage module, wherein the third commit request is associated with a second data; and
process the third commit request, wherein processing the third commit request causes a first copy of the second data to be stored in a slave partition of the persistent storage in the first storage module,
wherein the slave partition is associated with a master partition on the second storage module.
12. The storage appliance of claim 11 , wherein the first storage module is further configured to:
prior to receiving the third commit request:
receive a second copy of the second data from the client; and
temporarily store the second copy of the second data in a buffer in the first storage module.
13. The storage appliance of claim 10 , wherein processing the first commit request comprises:
storing the first commit request in a global submission queue, wherein the global submission queue maintains an order of all commit requests received by the first storage module,
wherein the first commit request is processed based on the order.
14. The storage appliance of claim 10 , wherein processing the first commit request comprises:
storing the first commit request in a client submission queue associated with the client, wherein the first commit request is associated with a global sequencing tag,
wherein the first commit request is processed based on an order specified by the global sequencing tag.
15. The storage appliance of claim 10 , wherein the fabric is a Peripheral Component Interconnect Express (PCIe) fabric.
16. The storage appliance of claim 10 , wherein the persistent storage in the first storage module comprises phase change memory.