Data storage method, SDN controller, and distributed network storage system
AssigneeHuawei Technologies Co., Ltd.
InventorJingcai Zhou
Life of this patent
Estimating the term…
Abstract
A data storage solution in a distributed network storage system, where a metadata management node sends a backup mapping relationship of a destination data node to a controller, where the controller generates a first flow entry according to the backup mapping relationship of the destination data node, and sends the first flow entry to a switch, where the first flow entry includes an address of the destination data node and an address of a backup data node of the destination data node, and the switch broadcasts, to the destination data node and the backup date node according to the first flow entry, a write request sent by a client.
Claims13
A distributed network storage system, comprising:
- a switch;
- a controller connected to the switch and comprising a memory and a processor coupled to the memory;
- data nodes;
- a metadata management node connected to the controller and communicates with the data nodes using the switch and configured to:
- receive a data storage request sent by a client;
- return metadata corresponding to to-be-written data to the client according to the data storage request, wherein the metadata comprises an address of a destination data node for storing the to-be-written data, wherein the destination data node is one of the data nodes;
- and
- send a backup mapping relationship of the destination data node to the controller, wherein the backup mapping relationship indicates a backup data node of the destination data node, and wherein the backup data node is one of the data nodes,
- wherein the controller is configured to:
- generate a first flow entry according to the backup mapping relationship of the destination data node sent by the metadata management node;
- and
- deliver the first flow entry to the switch, wherein the first flow entry comprises the address of the destination data node and an address of the backup data node of the destination data node,
- wherein the switch is configured to:
- receive a write request that is sent by the client according to the metadata, wherein the write request comprises the to-be-written data and the address of the destination data node;
- search a flow table according to the address of the destination data node for the first flow entry that matches the address of the destination data node;
- and
- broadcast the write request to the destination data node and the backup data node according to the first flow entry, and
- wherein the destination data node and the backup data node are separately configured to:
- receive the write request broadcast by the switch;
- and
- separately store the to-be-written data according to the write request.
A data storage method applied to a distributed network storage system, wherein the distributed network storage system comprises a metadata management node, data nodes, a controller, a switch, and a resource management node connected to the switch, and wherein the method comprises:
- receiving, by the controller, and from the metadata management node, a backup mapping relationship of a destination data node and a backup data node of the destination data node, wherein the destination data node is one of the data nodes, and wherein the backup data node is one of the data nodes;
- generating, by the controller, a first flow entry according to the backup mapping relationship, wherein the first flow entry comprises an address of the destination data node and an address of the backup data node, and wherein the first flow entry is used to broadcast a matching data packet to the destination data node and the backup data node;
- sending, by the controller, the first flow entry to the switch;
- receiving, by the controller, an address of the metadata management node sent by the resource management node;
- generating, by the controller, a second flow entry according to the address of the metadata management node, wherein the second flow entry comprises the address of the metadata management node, and wherein the second flow entry is used to add the matching data packet to a highest priority queue;
- and
- sending, by the controller, the second flow entry to the switch.
A controller in a distributed network storage system, wherein the distributed network storage system comprises a metadata management node, data nodes, the controller, and a switch, and wherein the controller comprises:
- an interface configured to receive a backup mapping relationship of a data node sent by the metadata management node;
- and
- a processor coupled to the interface and configured to:
- generate a first flow entry according to the backup mapping relationship of the data node;
- and
- send the first flow entry to the switch, wherein the first flow entry comprises an address of a destination data node and an address of a backup data node of the destination data node, wherein the first flow entry is used to broadcast a matching data packet to the destination data node and the backup data node, wherein the backup data node is one of the data nodes, and wherein the destination data node is one of the data nodes,
- wherein the interface is further configured to receive a changed backup mapping relationship of the destination data node sent by the metadata management node, and
- wherein the processor is further configured to:
- update the first flow entry according to the changed backup mapping relationship of the destination data node;
- and
- send an updated first flow entry to the switch.
Dependencies are inferred from the claim text.