JP2012053571A - Information processing unit, controller card and information processing unit control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the writing speed of a logical drive without worrying about coexistence of a plurality of file systems.
An information processing apparatus includes first setting means and second setting means. The first setting means sets the sector size of a logical drive composed of a plurality of storage devices to a positive integer multiple of the stripe group capacity capable of storing data other than the redundant code in one stripe group. The second setting means sets the cluster size to a positive integer multiple of the sector size set by the first setting means.
[Selection] Figure 3

Description

  Embodiments described herein relate generally to an information processing apparatus, a controller card, and a method for controlling the information processing apparatus that make data redundant and write the data in a plurality of storage devices.

  In general, a drive array device includes a drive array composed of a plurality of drives and a drive array control device that controls access to each drive in the drive array, and drives each drive in parallel to increase the access speed. It is known as an external storage device in which reliability is improved by a redundant configuration.

  The disk array control device generates parity data as data correction information for the write data transferred from the host device by exclusive OR operation and writes it to one of the plurality of drives. . This makes it possible to restore the failed drive data by performing an exclusive OR operation using the parity data and the remaining drive data for the failure of one of the multiple drives. Yes.

  A RAID (Redundant Array of Inexpensive Disks) method is known as one of such data redundancy methods using parity or the like. In the RAID system, the RAID level is classified into various levels in relation to the RAID data and the parity, and the representative level is level 5.

  When there is a write request less than the capacity of the stripe group, the RAID 5 logical drive needs to read existing data and parity that are insufficient to create parity before writing new data and parity. This is a factor that reduces the writing speed.

  In order to suppress a decrease in writing speed, by matching the data capacity per stripe group to the access unit of the file system, new data and parity can be written without reading existing data or parity. There is technology.

  However, with the technology described above, when multiple partitions are set for one logical drive and each partition is formatted with another file system, the capacity of the stripe group in the logical drive used in each file system is changed. It is necessary to deal with it individually. Since the access unit (cluster size) differs depending on the file system, it is necessary to change the stripe group capacity for each file system.

JP 2009-80549 A

  It is desired to improve the writing speed of a logical drive without worrying about the coexistence of a plurality of file systems.

  An object of the present invention is to provide an information processing apparatus, a controller card, and an information processing apparatus control method capable of improving the writing speed of a logical drive without worrying about coexistence of a plurality of file systems. It is in.

  According to the embodiment, a plurality of storage devices are managed in units of stripe groups, data is distributed and written to the plurality of storage devices, and the same among the plurality of first storage devices among the plurality of storage devices An information processing apparatus for storing a redundant code generated from data stored in a stripe group in the same stripe group of a second storage device of the plurality of storage devices includes: a first setting unit; Second setting means is provided. The first setting means sets the sector size of a logical drive composed of a plurality of storage devices to a positive integer multiple of the stripe group capacity capable of storing data other than the redundant code in one stripe group. The second setting means sets the cluster size to a positive integer multiple of the sector size set by the first setting means.

The block diagram which shows the system configuration | structure of the information processing apparatus of embodiment. The block diagram which shows the system configuration | structure of a RAID controller card. The figure which shows the typical data and parity arrangement | positioning example in the logical drive of RAID-5. The figure which shows the example which constructed | assembled RAID-5 with three drives and set the sector size of the logical drive to four stripe groups. The figure which shows the logical block address (LBA) set by the operating system at the time of constructing a logical drive by three drives shown in FIG. The block diagram which shows the structure for setting a logical sector size and a cluster size. The flowchart which shows the procedure which performs the setting of a logical sector size, and a cluster size.

  Hereinafter, embodiments will be described with reference to the drawings.

An information processing apparatus according to an embodiment will be described with reference to FIG. The information processing apparatus is realized as a computer.
FIG. 1 is a block diagram showing a system configuration of the computer 10. As shown in FIG. 1, the computer 10 includes a CPU 11, a north bridge 12, a main memory 13, a graphics controller 14, a VRAM 14A, a south bridge 16, a BIOS-ROM 17, a RAID controller card 18, a hard disk (HDD) 19, and the like. ing.

  The CPU 11 is a processor that controls the operation of each unit in the computer 10. The CPU 11 executes an operating system (file system) loaded from the HDD 19 to the main memory 13 and various programs that operate under the control of the operating system. The CPU 11 also executes a basic input / output system (BIOS) stored in the BIOS-ROM 17. Hereinafter, the basic input / output system itself stored in the BIOS-ROM 17 may be referred to as BIOS.

  The north bridge 12 is a bridge device that connects the local bus of the CPU 11 and the south bridge 16. The north bridge 12 has a function of executing communication with the graphics controller 14 via a bus, and also includes a memory controller that controls access to the main memory 13. The graphics controller 14 is a display controller that controls the display 15 on the main body side. The graphics controller 14 generates a video signal to be sent to the display 15 from the image data written in the VRAM 14A.

  The south bridge 16 is a controller that controls various devices of the PCI Express (PCIe) bus. The south bridge 16 is directly connected to the BIOS-ROM 17 and has a function of controlling them.

  RAID-5 is constructed by three hard disk drives 19A to 19C.

  Next, the system configuration of the RAID controller card 18 will be described with reference to the block diagram of FIG. The RAID controller card 18 includes a controller 301, a PCIe controller 302, a SAS / S-ATA controller 303, a cache memory 311, a nonvolatile memory 320, and the like.

  The controller 301 performs write processing and read processing related to RAID. In addition, when there is an inquiry about information related to the HDDs 19A to 19C from the south bridge 16, the controller 301 makes an answer to the inquiry. The PCIe controller 302 is a PCI Express interface for controlling transmission of data to and from the south bridge 16. The SAS / S-ATA controller 303 is an interface for controlling data transmission between the HDDs 19A to 19C. The cache memory 311 is configured by, for example, a DDR2 SDRAM (Double-Data-Rate 2 Synchronous Dynamic Random Access Memory). For example, when there is a write access to the HDDs 19 </ b> A to 19 </ b> C, the data transferred from the south bridge 16 is once written in the cache memory 311. The controller 301 calculates a parity (redundancy code) described later.

  The nonvolatile memory 320 stores RAID setting data 321, logical sector size data 322, an address conversion table 323, and the like. The RAID setting data 321 describes settings indicating information indicating the type of RAID, stripe size, and the like. The logical sector size data 322 describes a logical sector size that is a minimum access unit of a logical drive constituted by three hard disk drives 19A to 19C. In the address conversion table 323, the data address (data block address) on the disk array (disk device) 2 as viewed from the computer 10, that is, the logical address (logical block address) is stored in the physical data (data block) in the disk array 2. An address indicating the position of is described. The logical block address viewed from the computer 10 is determined based on the logical sector size. The file system executed by the computer 10 sets the cluster size, which is the minimum access unit from the file system, to a positive integer multiple of the logical sector size.

  FIG. 3 is an example of general data and parity arrangement in a RAID 5 logical drive constructed with N disks 0 to N-1. Each disk is divided and managed in units called stripes. For example, Data 0,0, Data 0,1,..., Data 0, N-2, and Parity 0 are stored in the stripe 0. In particular, a stripe in which user data is stored is called a data stripe, and a stripe in which parity is stored is called a parity stripe.

A group for creating one parity stripe from one stripe of each of N−1 disks is defined as a stripe group (SG). For example, SG (0) includes Data 0,0, Data 0,1,..., Data 0, N-2 that are data stripes, and Parity 0 that is a parity stripe. The data capacity S _SG of the stripe group excluding the parity stripe is S _DS × (N−1). Here, _SDS is the size of the data stripe and the parity stripe, that is, the stripe size.

  When the parity exists in the disk N-1, Parity m is redundant data of the data stripe generated by the following formula.

Parity m = (Data m, 0) XOR (Data m, 1) XOR… XOR (Data m, N-2)
Here, XOR indicates exclusive OR.

  For example, when data Data m, 0 is updated to new Data m, 0, the corresponding parity Parity m needs to be updated at the same time. The formula for creating a new new Parity m is:

(new Parity m) = (new Data m, 0) XOR (Data m, 1) XOR… XOR (Data m, N-2)
Or
(new Parity m) = (Data m, 0) XOR (new Data m, 0) XOR (Parity m)
In order to generate new Parity m, it is necessary to read the existing data Data m, 1 to Data m, N-2, or Data m, 0, Parity m from the drive, which has a serious adverse effect on the write speed of RAID5. ing.

  Therefore, in the RAID controller card 18, the sector size of the RAID 5 logical drive is set to n times the data capacity of the stripe group (n is a positive integer). By making the sector size n times the data capacity of the stripe group, all the data stripes of each stripe group are always updated, and it is not necessary to read existing data, so that the writing performance can be improved. .

  FIG. 4 shows an example in which a RAID 5 logical drive using three drives (HDDs 19A to 19C) is constructed by a RAID controller. Drive 0 corresponds to the HDD 19A, Drive 1 corresponds to the HDD 19B, and Drive 2 corresponds to the HDD 19C.

  In order to simplify the description, the stripe size is set to 512 bytes (one sector size of Drives 0 to 2). The sector size of the logical drive is set to a positive integer multiple of the stripe group capacity. The stripe group capacity is a size that can store data other than parity in one stripe group. In the example shown in FIG. 4, four stripe groups (SG0 to SG3, SG4 to SG7, etc.) are used as the sector size of the logical drive. Therefore, the sector size of the logical drive is 512 bytes × (3-1) × 4 = 4 KB (1 KB = 1024 bytes).

  When the file system acquires the sector size of the logical drive, access from the file system to the logical drive is performed in units of 4 KB. For this reason, in the writing process, all data stripes belonging to the four stripe groups are always updated, so reading for generating parity can be omitted, and writing performance can be improved.

  By the way, when the interface of the logical drive is SCSI (Small Computer System Interface), the controller 301 receives the ReadCapacity command transmitted from the operating system (file system), and thereby converts the individual information of the logical drive into the operating system (file system). ). In the individual information, information necessary for writing and reading data such as sector information including the sector size of the logical drive and the maximum address value is described.

  When the interface of the logical drive is ATA (Advanced Technology Attachment), the controller 301 receives the IDENTIFY DEVICE command and returns the individual information of the logical drive to the operating system (file system).

  FIG. 5 shows logical block addresses (LBA) set by the operating system when a logical drive is constructed by the three drives shown in FIG. In general, a logical block address is allocated for each 512-byte sector size which is the same as the sector size of a general hard disk drive. In this embodiment, a logical block address is assigned every 4 KB. Therefore, even a capacity exceeding 2 TB can be handled with a 32-bit address width.

  Next, a configuration for setting the logical sector size and the cluster size will be described with reference to the block diagram of FIG.

  The controller 301 includes a parity stripe number detection module 401, a stripe group capacity calculation module 402, a logical sector size creation module 403, a conversion table setting module 404, a transmission module 405, and the like.

  Based on the RAID setting data 321 stored in the non-volatile memory 320, the parity stripe number detection module 401 stores the number of parity stripes within one stripe group in RAID, that is, parity within one stripe group. Detect the number of drives. The parity stripe number detection module 401 detects the number of data stripes based on the type of RAID described in the RAID setting data. For example, if the RAID type is RAID-5, the number of parity stripes is 1. If the RAID type is RAID-6, the number of parity stripes is two.

  The stripe group capacity calculation module 402 can be stored in one stripe group based on the number of parity stripes detected by the parity stripe number detection module 401 and the total number of drives and stripe size described in the RAID setting data 321. The correct data capacity. The data here excludes parity.

The stripe group capacity S _SG is obtained by the following equation.

S _SG = S _DS × (km)
Here, _SDS is the stripe size, k is the total number of drives constituting the logical drive, and m is the number of drives in which parity in one stripe group is stored.

  The logical sector size creation module 403 sets the logical sector size of the logical drive based on the stripe capacity calculated by the stripe group capacity calculation module 402.

The sector size S _S is obtained by the following equation.

S _S = S _SG × n = S _DS × (km) × n
km is the number of data stripes in one stripe group. Therefore, the number of drives in which data in one stripe group is recorded in a distributed manner, in other words, the number of drives in which no parity is stored in one stripe group is described in the RAID setting data, and based on the number described. The sector size may be obtained.

  The logical sector size creation module 403 records the set logical sector size in the logical sector size data 322.

  The conversion table setting module 404 creates an address conversion table based on the logical sector size set by the logical sector size creation module 403. The conversion table setting module 404 stores the created address conversion table in the address conversion table 323. The address conversion table is used to convert a logical block address based on the logical sector size into a physical block address of a physical drive.

  The transmission module 405 receives the individual information including the logical sector size described in the logical sector size data 322 when the command for instructing the transmission of the individual information of the logical drive is transmitted from the operating system (file system) 410. File system) 410.

  The operating system (file system) 410 sets the cluster size to a positive integer multiple of the logical sector size included in the individual information.

  Next, the procedure for setting the logical sector size and the cluster size will be described with reference to the flowchart of FIG.

  The controller 301 sets the sector size of the logical drive (step 501). The operating system (file system) 410 requests the individual information of the logical drive from the controller 301 (step 502). In response to the request, the controller 301 transmits individual information of the logical drive including the logical sector size to the operating system (file system) 410 (step 503). The operating system (file system) 410 sets the cluster size based on the logical sector size included in the received individual information.

  With the above processing, the sector size of the logical drive can be set to a positive integer multiple of the stripe group capacity. In addition, the cluster size can be set to a positive integer multiple of the sector size of the logical drive. By setting the cluster size to a positive integer multiple of the sector size of the logical drive, all data stripes in one stripe group will be updated when performing write access from the operating system (file system). Since there is no need to read, the writing performance can be improved.

  In addition, by setting the logical drive sector size to a positive integer multiple of the stripe group capacity, the access unit (cluster size) will be a positive integer multiple of the sector size (minimum disk access unit) in any file system. Therefore, there is no need to worry about the coexistence of multiple file systems.

  Note that if the sector size of the logical drive you want to specify is not a positive integer multiple of the stripe group capacity, you can make the extra physical drive sector unused to make it a positive integer multiple of the stripe group capacity. is there. In this case, the capacity efficiency is lowered, but the writing performance can be improved.

  In this embodiment, the sector size of the logical drive is set to 4 times the stripe group capacity. However, the sector size of the logical drive is set to 1 time, 2 times, 3 times, 5 times, 6 times, etc. It may be set to a positive integer multiple of. In this embodiment, the cluster size is set to 1 times the sector size of the logical drive. However, the cluster size may be set to a positive integer multiple such as 2 times, 3 times, or 4 times the sector size of the logical drive. good.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  18 ... RAID controller card, 19A, 19B, 19C ... hard disk, 301 ... controller, 311 ... cache memory, 320 ... non-volatile memory, 321 ... RAID setting data, 322 ... logical sector size data, 323 ... address conversion table, 401 ... Parity stripe count detection module, 402 ... stripe group capacity calculation module, 403 ... logical sector size creation module, 404 ... conversion table setting module, 405 ... transmission module, 410 ... operating system.

Claims (5)

Multiple storage devices are managed in units of stripe groups, data is distributed and written to the plurality of storage devices, and stored in the same stripe group of a plurality of first storage devices among the plurality of storage devices A redundant code generated from the data being stored in the same stripe group of the second storage device of the plurality of storage devices,
First setting means for setting a sector size of a logical drive composed of the plurality of storage devices to a positive integer multiple of a stripe group capacity capable of storing data other than the redundant code in one stripe group;
An information processing apparatus comprising: a second setting unit that sets a cluster size to a positive integer multiple of the sector size set by the first setting unit. Storage means for storing the stripe size;
Obtaining means for obtaining the number of the plurality of first storage devices;
Calculating means for calculating the stripe group capacity based on the stripe size and the number of the plurality of first storage devices;
The information processing apparatus according to claim 1, wherein the first setting unit sets the sector size based on a stripe group capacity calculated by the calculation unit. The storage means stores first data indicating the number of the second storage devices and second data indicating the number of the plurality of storage devices,
The information processing apparatus according to claim 2, wherein the acquisition unit calculates the number of the plurality of storage devices based on the first data and the second data. Multiple storage devices are managed in units of stripe groups, data is distributed and written to the plurality of storage devices, and stored in the same stripe group of a plurality of first storage devices among the plurality of storage devices A controller card that stores a redundant code generated from the stored data in the same stripe group of a second storage device of the plurality of storage devices,
A positive integer multiple of the stripe group capacity capable of storing data other than the redundant code in one stripe group when the host requests transmission of the sector size of the logical drive composed of the plurality of storage devices. A controller card comprising transmitting means for transmitting the data as the sector size to the host. Multiple storage devices are managed in units of stripe groups, data is distributed and written to the plurality of storage devices, and stored in the same stripe group of a plurality of first storage devices among the plurality of storage devices A method of controlling an information processing apparatus that stores a redundant code generated from data stored in the same stripe group of a second storage device of the plurality of storage devices,
The sector size of the logical drive composed of the plurality of storage devices is set to a positive integer multiple of the stripe group capacity capable of storing data other than the redundant code in one stripe group,
A method for controlling an information processing apparatus, wherein a cluster size is set to a positive integer multiple of the set sector size.

Images