JP2023183437A - Memory system and memory method - Google Patents

Abstract

To provide a memory system and a memory method for protecting a solid state drive.SOLUTION: A memory system includes two SSDs 18, 19 and a controller 15 for controlling the SSDs 18, 19. The controller 15 controls, in writing data into an SSD, the SSD 19 of the two SSDs 18, 19 to be "inactive" that prohibits read, write and internal processing, while writing data A into the SSD 18. The data A is stored in a buffer unit 16. Thereafter, the controller controls the SSD 18 to be "inactive" so as to prohibit read, write and internal processing, while writing the data A into the SSD 19.SELECTED DRAWING: Figure 2

Description

The present invention relates to a storage system and a storage method.

As background technology in this technical field, there is Japanese Patent Application Laid-Open No. 10-3434 (Patent Document 1). This publication states, ``A nonvolatile register is provided to store write data, a write destination address, and a flag indicating that writing is in progress, and the write data is written to the buffer memory and simultaneously written to the nonvolatile register. When writing to the non-volatile register is completed, the flag indicating that the semiconductor disk device is in a busy state is cleared, and even if the power is cut off during writing to the flash memory array, the non-volatile register will continue to operate. It is possible to guarantee the written data (see summary).

Another background technology is Japanese Patent Application Publication No. 2010-102369 (Patent Document 2). This bulletin states, ``The processor sets the SSD 352A in a writable state, sets the SSD 352B, which can acquire the same data, in a writable state, and sets the SSD 352A in a writable state. , writes predetermined data in the cache memory, receives a data read request from the host computer, and when the data storage destination is the SSD 352A that is set as writable, the SSD 352B writes the specified data to the read request from the SSD 352B. It is configured to acquire data and send the acquired data to the host computer.'' (see summary).

Japanese Patent Application Publication No. 10-3434 Japanese Patent Application Publication No. 2010-102369

However, the techniques disclosed in Patent Documents 1 and 2 have room for improvement from the viewpoint of protecting storage devices such as solid state drives.
Therefore, an object of the present invention is to provide a storage system and a storage method that can protect solid state drives.

In order to solve the above problems, the present invention includes a plurality of solid state drives and a control unit that controls each of the solid state drives, and the control unit controls the solid state drive. When reading or writing data, some of the solid state drives are suspended so that they cannot read, write, or perform internal processing, and the remaining solid state drives are suspended. The method is characterized in that the data is read from or written to a solid state drive.

According to the present invention, it is possible to provide a storage system and a storage method that can protect solid state drives.
Problems, configurations, and effects other than those described above will be made clear by the following description of the embodiments.

FIG. 1 is a block diagram showing the system configuration of a storage system according to an embodiment of the present invention. 1 is a block diagram showing a system configuration centered on a controller in a storage system according to an embodiment of the present invention. FIG. FIG. 2 is a block diagram illustrating a program in a storage system according to an embodiment of the present invention. 1 is a block diagram showing a schematic configuration of an SSD in a storage system according to an embodiment of the present invention. FIG. 1 is a block diagram showing a system configuration centered on a controller in a storage system according to an embodiment of the present invention. FIG.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing the system configuration of a storage system according to an embodiment of the present invention. The storage system 1 is, for example, an auxiliary storage system mounted on a railway vehicle. In the storage system 1, a CPU (Central Processing Unit) 11 performs various calculations and centrally controls each part of the storage system 1. A main memory 20 and an I/O (Input/Output) device 13 are connected to the CPU 11 . The main memory 20 is a storage device that serves as a work area for the CPU 11. The I/O device 13 inputs data and signals to the CPU 11 from the outside, and outputs them to the outside.

A controller 15 and a communication section 14 are connected to the I/O device 13 . The communication unit 14 is a communication interface through which the storage system 1 communicates with the outside. Connected to the controller 15 are a plurality of solid state drives (SSD) 18 and 19, which are two auxiliary storage devices in this example, and a buffer unit 16 made up of volatile memory. has been done. The controller 15 controls the operations of the SSDs 18 and 19 and the buffer unit 16. The same program 40 is set up in each SSD 18, 19. The CPU 11 reads the program 40 into the main memory 20 and executes processing based on the program 40.

FIG. 2 is a block diagram showing the system configuration centered on the controller. The controller 15 includes an auxiliary storage device input/output control section 21 that controls the SSDs 18 and 19 and a buffer control section 17 that controls the buffer section 16. A register 30 is provided in the buffer section 16 .

FIG. 3 is a block diagram illustrating the program 40. As shown in FIG. The program 40 includes an application program 41, a disk driver 42, and an OS (Operating System) 43. The OS 43 is basic software that performs resource management, process management, timer management, etc., for example. The disk driver 42 is a module that issues write commands and read commands to the SSDs 18 and 19. The application program 41 is a module for performing program processing for using the storage system 1 (computer system) for various purposes. The application program 41 is, for example, a railway operation management system if the storage system 1 is an auxiliary storage system mounted on a railway vehicle. Applications that use the computer system (storage system 1) for various purposes include, for example, a railway operation management system, a power system control system, a gas management system, a road management system, and the like.

By the way, if the storage system 1 is an auxiliary storage system mounted on a railway vehicle, the storage system 1 will be exposed to vibrations, so an SSD may be used as the auxiliary storage device instead of a magnetic storage device. is common.

However, the auxiliary storage system installed in a railway vehicle cannot be properly shut down by the vehicle driver or the like. That is, the auxiliary storage system installed in a railway vehicle operates in a state where it cannot be operated at all by the driver or the like. Then, when the main power of the vehicle is turned on, the power of the auxiliary storage system is also turned on, and when the main power of the vehicle is turned off, the power of the auxiliary storage system is also turned off.

Therefore, there is a possibility that the SSD's power is suddenly cut off while data is being read or written to the SSD's flash memory, and in that case, the logical-physical conversion table, which is the SSD's management information, may be damaged. When this happens, a problem arises in that the SSD cannot be recognized from the host side. This point will be explained in more detail.

FIG. 4 is a block diagram showing a schematic configuration of the SSDs 18 and 19. The SSDs 18 and 19 include a dynamic random access memory (DRAM) 51 that is a volatile memory, a NAND flash memory 52 that is a nonvolatile memory, and a controller 53 that controls the SSD. The data 55 is data to be stored in the SSD, and the logical-physical conversion table 56 is management information for the SSD, both of which are stored in the flash memory 52.

When reading the data 55 from the SSDs 18 and 19, the data 55 is read from the cache of the DRAM 51, or the data 55 is read from the flash memory 52 by referring to the logical-physical conversion table 56. If the power of the SSD suddenly turns off during the read operation of the data 55, the logical-physical conversion table 56 will fail. However, the SSD may have a function to repair the logical-physical conversion table 56 that has failed after the power is restored. However, when repairing, the failed logical-physical conversion table 56 is attempted to be repaired from the logical-physical conversion table 56 on the flash memory 52, but the logical-physical conversion table 56 may be updated with an abnormal value. In that case, the logical-physical conversion table 56 becomes abnormal and the data 55 cannot be read from the SSD.

Even if the power to the SSD is suddenly cut off at the timing of writing data 55 to the SSDs 18 and 19, the failed logical-physical conversion table will be detected from the logical-physical conversion table 56 on the flash memory 52 after the power to the SSD is restored. However, the logical-physical conversion table 56 may be updated with an abnormal value, and in that case, the data 55 cannot be written to the SSD.

In contrast, the technique disclosed in Patent Document 1 attempts to avoid problems when writing data to a flash memory using registers. That is, the problem is avoided by providing a non-volatile register and storing the write data, write destination address, and writing flag in the register. If the power is cut off while writing from the buffer memory to the flash memory, the nonvolatile registers are referenced during self-diagnosis at the next startup, and if the writing flag is held, the information stored in the registers is The data being written is guaranteed by restarting the write process based on the data being written.

However, even with the technique disclosed in Patent Document 1, the logical-physical conversion table may be damaged due to a sudden power cut during data writing, and in this case, the flash memory becomes unrecognizable from the host side.

Further, in the technology of Patent Document 2, RAID (Redundant Arrays of Inexpensive Disks) technology is used for two SSDs. However, even with the technique disclosed in Patent Document 2, the logical-physical conversion table may be damaged due to sudden power interruption during data reading or writing, and in this case, the SSD becomes unrecognizable from the host side.

Therefore, technical improvements in the storage system 1 that solve the above-mentioned problems will be described below.
The operating states of the SSDs 18 and 19 may be controlled by the controller 15 or by commands issued by the CPU 11 based on the disk driver 42. That is, in the embodiment of the present invention, the "control unit" is executed by the CPU 11 based on the controller 15 or the disk driver 42. In the following, an example will be explained in which the operating states of the SSDs 18 and 19 are controlled by the controller 15.

First, referring to FIG. 2, the controller 15 places a part of the plurality of SSDs, in this example, the SSD 18, in an "operating state", and puts the rest of the plurality of SSDs, in this example, the SSD 19, in a "dormant state". . The "operating state" is a state in which the power of the SSD in the operating state is fully turned on, the SSD can read and write data, and the internal processing of the SSD can be performed completely. The "dormant state" is a state in which the SSD is completely powered off, data cannot be read or written to the SSD, and internal processing of the SSD cannot be performed.

In this way, the controller 15 puts the SSD 18 in the "active state" and the SSD 19 in the "dormant state," and then sends the data A to be written from the host side via the auxiliary storage device input/output control unit 21 to the "active state." Write to the SSD 18 in “status”. On the other hand, the controller 15 also writes the same data A to the buffer section 16 via the buffer control section 17. After completion of this write, the controller 15 writes a flag to the register 30 in the buffer section 16, indicating that the write data (data A) is stored in the buffer section 16.

Next, the controller 15 performs matching processing of the stored data between some of the plurality of SSDs and the other remaining parts (SSD 18 and SSD 19 in this example). As shown in FIG. 5, first, the controller 15 puts the SSD 18, which was in the "operating state" in the example of FIG. The state transition is performed as follows.

As described above, since the write data (data A) from the host side is stored in the buffer section 16, this data A is used to perform the data matching process. When starting the data matching process, a flag indicating that the data matching process is in progress is written to the register 30 in the buffer 16 section. While the flag indicating that data matching processing is in progress is set, the write data (data B) from the host side is stored in a separate area from the storage area for data A in the buffer unit 16. Furthermore, when data B is written to a separate area of the buffer unit 16, a flag indicating that the data is stored in a separate area from the storage area of data A of the buffer unit 16 is sent to the register 30 in the buffer unit 16. It is also possible to write data to the host, or the host can be configured to prohibit writing data while the data matching process is in progress.

As described above, after the SSD 18 which was in the "active state" is brought into the "dormant state" and the SSD 19 which was in the "dormant state" is brought into the "active state", the following processing is performed. That is, the controller 15 writes the data A stored in the buffer section 16 to the SSD 19 in the operating state. After completion of this write, the controller 15 erases the flag in the register 30 indicating that the write data (data A) is stored. If the flag indicating that the data (data B) is stored in a separate area from the data A storage area of the buffer section 16 in the register 30 is not set, the data matching process is completed here. .

If a flag indicating that the data (data B) is stored in a separate area from the data A storage area of the buffer section 16 is set in the register 30, the controller 15 saves the data A of the buffer section 16. Data matching processing is performed using data B stored in a separate area from the storage area of . That is, after writing data B to the SSD 19 in the "operating state," the controller 15 changes the state of the SSD 19 from the "active state" to the "dormant state," and changes the other SSD 18 from the "dormant state" to the "active state." state. Then, data B is written again to the SSD 18 in the "operating state". When the processing up to this point is completed, a flag indicating that the data in the register 30 is being matched and a flag indicating that data B is stored in a separate area from the storage area of data A in the buffer unit 16 are set. to erase.

Note that since the buffer section 16 is a volatile memory, the data in the buffer section 16 is cleared at the timing when the power of the storage system 1 is turned off. The data matching process described above may be performed at any timing or at regular intervals.

As shown in Figure 2, if the power to the storage system 1 is suddenly cut off during the process of writing data A to the SSD 18 in the "operating state", the SSD 18 may become unrecognizable from the host side. There is sex. However, since the SSD 19 was in the "hibernation state" at the time of the power outage, it was not affected by the power outage, and if the SSD 19 was then put into the "operating state", data could be read or written to the SSD 19. is possible, and various data stored in the SSD 19 is also preserved. Note that in the case of mirroring, since both are in the operating state, they will be affected by a power outage, but if they are in the "dormant state" as in this embodiment, they will not be affected by the power outage.

Furthermore, as shown in FIG. 5, if the power to the storage system 1 is suddenly cut off while data A is being written to the SSD 19 in the "operating state", the SSD 19 may become unrecognizable from the host side after that. . However, since the SSD 18 was in a "dormant state" at the time of the power outage, it was not affected by the power outage, and if the SSD 18 was subsequently put into an "operating state", data could be read or written to the SSD 18. is possible, and the various data stored in the SSD 18, including data A, is preserved.

Therefore, according to the storage system 1, it is possible to protect the SSD.
Furthermore, since the storage system 1 performs data matching processing as described above, all data written to the storage system 1 is held in both the SSD 18 and the SSD 19. Therefore, even if one of the SSDs 18 and 19 becomes unrecognizable from the host side as described above due to a sudden power outage, all the data written to the storage system 1 will be Data can be used.

Further, the above-described processing is also executed when reading data. That is, before reading data, one of the SSDs 18 and 19 is placed in an "active state" and the other is placed in a "dormant state", and data is read from the SSD in the "active state". Even if the power to the storage system 1 is suddenly cut off while this data is being read, the SSD that has been put into the "hibernation state" will operate normally by putting it into the "operating state", so the SSD can be protected.

The present invention has been described as a storage system (auxiliary storage system/device) and a storage method (method for protecting an auxiliary storage device in an onboard device) in an onboard device for use in a railway vehicle. However, the present invention is not limited to the embodiments described above, and includes various modifications. For example, the above-described embodiments have been described in detail to explain the present invention in an easy-to-understand manner, and the present invention is not necessarily limited to having all the configurations described.

Further, the control lines and information lines are shown to be necessary for explanation purposes, and not all control lines and information lines are necessarily shown in the product. In reality, almost all configurations can be considered to be interconnected.

1 Storage system 11 CPU (control unit)
15 Control controller (control unit)
16 Buffer section 18 Solid state drive (SSD)
19 Solid State Drive (SSD)

Claims (4)

multiple solid state drives,
a control unit that controls each of the solid state drives,
When reading or writing data to the solid state drive, the control unit may cause some of the solid state drives among the plurality of solid state drives to read or write data internally. A storage system characterized in that the data is read from or written to the remaining solid state drives while being suspended so that no processing can be performed. It further includes a buffer section for buffering data,
When writing the data to the solid state drive, the control unit also buffers the data to the buffer unit, and then writes some of the solid state drives to which the data has been written.・Hibernate the state drive so that it cannot read, write, or perform internal processing, restart the remaining suspended solid state drive so that it can read, write, or perform internal processing, and then 2. The storage system according to claim 1, further comprising: writing the data buffered by the buffer section to a state drive. For a storage system that reads or writes data to multiple solid state drives, when reading or writing data to the solid state drives, multiple solid state drives - Some of the solid state drives among the drives are suspended so that they cannot be read, written or internally processed, and the data is read or written to the remaining solid state drives; A memory method characterized by For the storage system further comprising a buffer unit for buffering data, when writing the data to the solid state drive, the data is also buffered to the buffer unit, and then , some of the solid-state drives to which the data has been written are suspended so that they cannot be read, written, or internally processed, and the rest of the solid-state drives that have been suspended are suspended from reading, writing, or internal processing. 4. The storage method according to claim 3, further comprising the step of resuming operation to enable processing and writing the data buffered by the buffer section to the remaining solid state drive.

Images