Scale-out storage system and configuration information control method for implementing high-availability, high-speed failover

1. A scale-out-type storage system in which a cluster is constructed by linking a plurality of nodes, at least two or more nodes among the plurality of nodes each comprising:

a cluster controller which controls processing spanning the whole cluster;

a node controller which performs closed processing control on its own node;

a plurality of subcluster processes which are processes which execute I/O processing in their own node, which form a subcluster between processes in their own node, and which are synchronized with work-type/standby-type corresponding processes in the other nodes; and

a nonvolatile data store which is shared by the whole cluster,

wherein the data store holds, as global configuration information, configuration information which includes information that must be shared by the whole cluster among the configuration information of the storage system,

wherein the subcluster processes hold, as local configuration information, configuration information which is required for their own subcluster process to operate among the configuration information of the storage system, and

wherein a work-type subcluster process is capable of executing the I/O processing without accessing the data store,

wherein the cluster controller is configured to:

receive an update request,

update a portion of the global configuration information based on the update request,

acquire an identifier for the portion of the global configuration information that is updated and acquire data based on the identifier,

specify a subcluster identifier based on the acquired data and acquire a subcluster, the acquired subcluster having the same subcluster identifier as the subcluster which is specified,

specify a node based on the acquired subcluster, the specified node is where the acquired subcluster is stored,

send a local configuration information update request to the node controller, such that the node controller updates the local configuration information corresponding to the update request.

2. The storage system according to claim 1 ,

wherein a standby-type subcluster process which corresponds to the subcluster process in one of the other nodes is switched to a working-type when the node which comprises the working-type subcluster process which is executing the I/O processing has entered a stopped or non-working state, and

wherein the subcluster process which has been switched to the working-type inherits the execution of the I/O processing by using the local configuration information.

3. The storage system according to claim 1 ,

wherein, when the global configuration information is updated, specific local configuration information which includes information corresponding to the update request among a plurality of the local configuration information is updated in response to the update to the global configuration information.

4. The storage system according to claim 3 ,

wherein the global configuration information contains control information which makes it possible to identify whether each of the plurality of subcluster processes is of a working type or a standby type,

wherein, when the update request to the global configuration information has been requested, the cluster controller updates the global configuration information and specifies the work-type subcluster process holding the specific local configuration information based on the control information, and

wherein the node controller of the node, which comprises the work-type subcluster process that is specified, updates the specific local configuration information by means of the global configuration information update request.

5. A configuration information control method of controlling configuration information of a scale-out-type storage system in which a cluster is constructed by linking a plurality of nodes, at least two or more nodes among the plurality of nodes in the storage system each comprising:

a cluster controller which controls processing spanning the whole cluster;

a node controller which performs closed processing control on its own node;

a plurality of subcluster processes which are processes which execute I/O processing in their own node, which form a subcluster between processes in their own node, and which are synchronized with work-type/standby-type corresponding processes in the other nodes; and

a nonvolatile data store which is shared by the whole cluster,

wherein the data store holds, as global configuration information, configuration information which includes information that must be shared by the whole cluster among the configuration information of the storage system,

wherein the subcluster processes hold, as local configuration information, configuration information which is required for their own subcluster process to operate among the configuration information of the storage system, and

wherein the work-type subcluster process is capable of executing the I/O processing without accessing the data store,

the cluster controller is configured to:

receive an update request,

update a portion of the global configuration information based on the update request,

acquire an identifier for the portion of the global configuration information that is updated and acquire data based on the identifier,

specify a subcluster identifier based on the acquired data and acquire a subcluster, the acquired subcluster having the same subcluster identifier as the subcluster which is specified,

specify a node based on the acquired subcluster, the specified node is where the acquired subcluster is stored,

send a local configuration information update request to the node controller, such that the node controller updates the local configuration information corresponding to the update request.

6. The configuration information control method according to claim 5 , comprising:

switching the standby-type subcluster process which corresponds to the subcluster process in one of the other nodes to a working-type when the node which comprises the work-type subcluster process which is executing the I/O processing has entered a stopped or non-working state, and

causing the subcluster process which has been switched to the working-type in the switching step to inherits the execution of the I/O processing by using the local configuration information.

7. The configuration information control method according to claim 5 ,

wherein, when the global configuration information is updated, specific local configuration information which includes information corresponding to the request among a plurality of the local configuration information is updated in response to the update to the global configuration information.

8. The configuration information control method according to claim 7 , wherein the method comprises:

the global configuration information containing control information which makes it possible to identify whether each of the plurality of subcluster processes is of a working type or a standby type, the configuration information control method further comprising:

updating, via the cluster controller, the global configuration information when the update request has been requested;

specifying, via the cluster controller, the work-type subcluster process holding the specific local configuration information based on the control information; and

updating the node controller of the node, which comprises the work-type subcluster process that is specified, to the specific local configuration information by means of the global configuration information update request.