Multi-node storage system and cooperation method for performing input output
There is provided a storage system that can avoid an increase in inter-node communication in the cooperation between file service and block service. The storage system includes a management section that manages a file processing section of a node in a currently active system and a file processing section of a node in a standby system as a pair, sets the file processing section of the node in the currently active system to be operational, manages a block processing section of the node in the currently active system and a block processing section of the node in the standby system as a pair, and sets the block processing section of the node in the currently active system to be operational.
1. A storage system in which a plurality of nodes that provide file service for performing I/O (input/output) on a file basis and block service for performing I/O on a block basis are coupled with each other via a network,
wherein the plurality of nodes each include a storage device that stores data, a file processing section that accepts a file I/O request from a file client and converts the file I/O request into a block I/O request in the file service, and a block processing section that carries out a process for performing I/O from/to the storage device based on the block I/O request in the block service,
the plurality of nodes include a node in a currently active system and a node in a standby system,
the storage system further include a management section that manages the file processing section of the node in the currently active system and the file processing section of the node in the standby system as a pair, sets the file processing section of the node in the currently active system to be operational, manages the block processing section of the node in the currently active system and the block processing section of the node in the standby system as a pair, and sets the block processing section of the node in the currently active system to be operational;
wherein the node in the currently active system is formed of a plurality of nodes,
the node in the standby system is provided with the file processing sections and the block processing sections that are respectively paired with the file processing sections and the block processing sections of the plurality of nodes in the currently active system; and
wherein the management section manages node state information representing states of the nodes, when failure occurs in any of the plurality of nodes in the currently active system, the management section stores information representing that the node is blocked in the node state information, and refers to the node state information, when the number of failed nodes is one, the management section sets the file processing section of the node in the standby system that is paired with the file processing section of the failed node to be operational, and sets the block processing section of the node in the standby system that is paired with the block processing section of the failed node to be operational.
2. The storage system according to claim 1 ,
wherein the number of hardware resources for the node in the standby system is greater than the number of hardware resources for any of the plurality of nodes in the currently active system.
3. The storage system according to claim 1 ,
wherein in the node in the standby system, when failure occurs in any of the plurality of nodes in the currently active system, a memory is allocated to the file processing section paired with the file processing section of the failed node.
4. The storage system according to claim 1 ,
wherein the management section refers to the node state information, when the number of failed nodes is two, the management section instructs the file processing sections and the block processing sections of the nodes where no failure has occurred, out of the plurality of nodes in the currently active system, not to perform the process, and instructs the file processing sections and the block processing sections, which are during operation, of the node in the standby system not to perform the process.
5. The storage system according to claim 1 ,
wherein when accepting a file I/O request from the file client, the file processing section of the node in the currently active system converts the file I/O request into a block I/O request,
the block processing section of the node in the currently active system carries out the process for performing I/O from/to the storage device of the node in the currently active system based on the block I/O request, and instructs the block processing section of the node in the standby system that is paired with the block processing section of the node in the currently active system to make data redundant so as to perform I/O from/to the storage device of the node in the standby system based on the block I/O request, and
when failure occurs in the node in the standby system, the management section instructs the block processing section of the node in the currently active system to stop making the data redundant.
6. A cooperation method for causing file service for performing I/O (input/output) on a file basis and block service for performing I/O on a block basis to cooperate with each other in a storage system in which a plurality of nodes that provide the file service and the block service are coupled with each other via a network,
wherein the plurality of nodes each include a storage device that stores data, a file processing section that accepts a file I/O request from a file client and converts the file I/O request into a block I/O request in the file service, and a block processing section that carries out a process for performing I/O from/to the storage device based on the block I/O request in the block service,
the plurality of nodes include a node in a currently active system and a node in a standby system, and
the method comprising, by a management section,
managing the file processing section of the node in the currently active system and the file processing section of the node in the standby system as a pair,
setting the file processing section of the node in the currently active system to be operational,
managing the block processing section of the node in the currently active system and the block processing section of the node in the standby system as a pair, and
setting the block processing section of the node in the currently active system to be operational,
wherein the node in the currently active system is formed of a plurality of nodes,
the node in the standby system is provided with the file processing sections and the block processing sections that are respectively paired with the file processing sections and the block processing sections of the plurality of nodes in the currently active system; and further comprising:
the management section
managing node state information representing states of the nodes, when failure occurs in any of the plurality of nodes in the currently active system,
storing information representing that the node is blocked in the node state information,
referring to the node state information, when the number of failed nodes is one,
setting the file processing section of the node in the standby system that is paired with the file processing section of the failed node to be operational, and
setting the block processing section of the node in the standby system that is paired with the block processing section of the failed node to be operational.
7. A non-transitory computer readable medium storing a program for providing file service for performing I/O (input/output) on a file basis and block service for performing I/O on a block basis in a storage system in which a plurality of nodes that provide the file service and the block service are coupled with each other via a network,
wherein the plurality of nodes each include a storage device that stores data, a file processing section that accepts a file I/O request from a file client and converts the file I/O request into a block I/O request in the file service, and a block processing section that carries out a process for performing I/O from/to the storage device based on the block I/O request in the block service,
the plurality of nodes include a node in a currently active system and a node in a standby system,
the program causing a computer to carry out a process of
managing the file processing section of the node in the currently active system and the file processing section of the node in the standby system as a pair, setting the file processing section of the node in the currently active system to be operational,
managing the block processing section of the node in the currently active system and the block processing section of the node in the standby system as a pair, and
setting the block processing section of the node in the currently active system to be operational,
wherein the node in the currently active system is formed of a plurality of nodes,
the node in the standby system is provided with the file processing sections and the block processing sections that are respectively paired with the file processing sections and the block processing sections of the plurality of nodes in the currently active system; and the process further comprising
managing node state information representing states of the nodes, when failure occurs in any of the plurality of nodes in the currently active system,
storing information representing that the node is blocked in the node state information,
referring to the node state information, when the number of failed nodes is one,
setting the file processing section of the node in the standby system that is paired with the file processing section of the failed node to be operational, and
setting the block processing section of the node in the standby system that is paired with the block processing section of the failed node to be operational.