CN107832168B

CN107832168B - Solid state disk data protection method

Info

Publication number: CN107832168B
Application number: CN201710950534.4A
Authority: CN
Inventors: 王猛; 徐伟华
Original assignee: Ramaxel Technology Shenzhen Co Ltd
Current assignee: Ramaxel Technology Shenzhen Co Ltd
Priority date: 2017-10-13
Filing date: 2017-10-13
Publication date: 2020-10-16
Anticipated expiration: 2037-10-13
Also published as: CN107832168A

Abstract

The invention discloses a method for protecting solid state disk data, which comprises the steps of respectively selecting N blocks in each channel to form a transverse RAID stripe group, wherein at least 1 Block is a data check Block, and N is greater than 1; the method is characterized in that the blocks belonging to the same transverse RAID stripe group and belonging to the same channel are at least separated by more than 1 block, preferably 512 blocks. The invention provides the optimized RAID strip composition especially aiming at the characteristic that the 3D TLC is easy to have errors spread to adjacent physical blocks, greatly reduces the probability that a plurality of block data errors occur in the same strip at the same time, and improves the safety performance of data.

Description

Solid state disk data protection method

Technical Field

The invention relates to a solid state disk control technology, in particular to a method for protecting data of a solid state disk.

Background

FIG. 1 is a typical NAND composition schematic: DIE, a unit that can operate independently and concurrently, called a channel; a Block, which is an independently erasable unit, wherein after data at each physical position is written, the whole Block must be erased before the next writing; page, read-write-unit, pages within the same physical block must be programmed in order 0- >1- >2- >3 ….

Fig. 2 is a schematic diagram of a typical 7+1RAID protection policy, which may include multiple physical planes from each independent concurrent operation unit, and one or more physical blocks (one will be described later) are allocated from each Plane to form a Stripe Group RAID Stripe Group. Taking the 7+1RAID protection policy as an example, performing exclusive or on every 7 pages of data in each stripe group, and writing a result into a physical page corresponding to the last physical block, which is called Parity. When an error occurs in one of the physical blocks, the error can be corrected by other 7 pages of data in the stripe.

Fig. 3 is a schematic diagram of a NANDFLASH error scenario of 3 DTLC; when an error occurs in a physical block, for example, physical block4 of Plane0, there is a large probability that data corruption will occur in an adjacent physical block, for example, physical block5 of Plane1, which falls within the same DIE. In the 7+1RAID component, because the allocation of the RAID physical blocks is that each Plane is randomly allocated according to the currently available blank block, there is a certain probability that the physical blocks of the same DIE and different planes in a certain Stripe Group are consecutive, and when an error occurs in this type of RAID Stripe Group, because there are more than 1 error in the same Stripe, it is not feasible to correct the RAID error.

Disclosure of Invention

Aiming at the defects, the invention aims to improve the existing stripe group grouping strategy, reduce the probability of the simultaneous damage of a plurality of pages of data in the same stripe group and further realize the aim of improving the data security.

In order to solve the above problems, the invention provides a method for protecting solid state disk data, wherein N blocks are respectively selected from each channel to form a transverse RAID stripe group, wherein at least 1 Block is a data check Block, N > 1; the method is characterized in that the blocks belonging to the same transverse RAID stripe group and belonging to the same channel are at least separated by more than 1 block, preferably 512 blocks.

The method for protecting the data of the solid state disk is characterized in that logical planes are respectively established for physical planes except a physical Plane0 in the same channel; performing dislocation recombination on the sequence of the block sequence numbers of the logical planes and the sequence of the blocks of the corresponding physical planes, specifically performing recombination through a block logical mapping table, mapping the blocks with the sequence numbers of P1+ M1 in the physical planes to the blocks with the sequence numbers of P1 in the logical planes, if the number of the planes is more than 2, taking different values for M1 selected by different planes, taking M1 as 512, and when a transverse RAID stripe group is formed, allocating the physical Plane0 to the blocks of the same channel for grouping according to the block sequence numbers of the physical Plane 0; the physical planes other than the physical Plane0 are assigned with the block sequence numbers of the logical planes corresponding thereto.

The method for protecting the data of the solid state disk is characterized in that when the free space of the solid state disk is large, scanning is carried out by taking block as a unit in a background when a solid state disk controller is free, and when data errors occur in a certain block when the data are read, the data are recovered through a transverse RAID stripe group.

The invention provides the optimized RAID stripe composition particularly aiming at the characteristic that 3DTLC is easy to cause error diffusion to adjacent physical blocks, greatly reduces the probability of a plurality of block data errors occurring in the same stripe at the same time, and improves the safety performance of data.

Drawings

FIG. 1 is a typical NAND composition schematic;

FIG. 2 is a diagram of an exemplary 7+1RAID protection strategy;

fig. 3 is a schematic diagram of a NANDFLASH error scenario of 3 DTLC;

FIG. 4 is a schematic diagram of an improved RAID stripe composition;

fig. 5 is a schematic flow chart of the specific implementation.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Particularly, aiming at the characteristic that 3DTLC is easy to cause error diffusion to adjacent physical blocks, the composition of RAID stripes is optimized so as to reduce the probability of multiple page data errors occurring in the same stripe at the same time. Respectively selecting N blocks in each channel to form a transverse RAID stripe group, wherein at least 1 Block is a data check Block, N > 1; in order to reduce the probability of multiple page data errors occurring in the same stripe, it is necessary to ensure that blocks belonging to the same horizontal RAID stripe group are not adjacent.

FIG. 4 is a schematic diagram of an improved RAID stripe composition; to ensure that blocks of the same horizontal RAID stripe group are not adjacent, the convenience of actual allocation is also taken into account. A conventional same channel is made up of more than 2 physical planes, each physical plane being made up of a plurality of blocks. Take 2 physical planes as an example: physical Plane0 is composed of Block0, Block2, Block4.. physical Plane1 is composed of Block1, Block3, Block 5.; when the conventional stripe is distributed, the same number of blocks are taken from the same position of each physical plane in each channel to form a stripe group, and if the stripe group is formed according to the conventional physical planes, the blocks belonging to the same channel in the same stripe group are adjacent inevitably; respectively establishing logical planes for physical planes except the physical Plane0 in the same channel; performing dislocation recombination on the block sequence formed by the logic planes and the block sequence of the corresponding physical plane, specifically performing recombination through a block logic mapping table; blocks with sequence number P1+ M1 in the physical plane are mapped to blocks with sequence number P1 in the logical plane, and if the number of planes is greater than 2, then M1 for different plane selections takes different values. Take 2 planes as an example, take M1 as 512. When the horizontal RAID stripe group is distributed, the block distribution of the same channel is distributed according to the block sequence numbers of the physical Plane0 and the logical Plane, for example, M1 is 512, the block interval of the same channel of the same horizontal RAID stripe group is 512, so that the situation that the blocks of the same stripe group belonging to the same channel are not adjacent to each other necessarily is ensured, and the probability that a plurality of page data errors occur in the same stripe at the same time is reduced. Because the physical block intervals of different Plane planes in the same channel DIE in any horizontal RAID stripe group RAIDGroup are large, the failure of the RAID mechanism caused by the damage of the physical block of a certain Plane can be avoided.

As shown in FIG. 4, when the Plane0 physical block4 of DIE0 is damaged, it will destroy the Plane1 physical block5, since these two physical blocks are respectively in RAIDgroup2/258 at this time, and only one data in the corresponding Group is damaged, so that the correct data can be recovered through the respective stripes.

Fig. 5 is a schematic diagram of a specific implementation flow, which mainly relates to an algorithm management module and a NAND driver module:

in an algorithm module, continuous logic BANK is adopted to abstract a physical block set in an independent concurrency unit;

physical blocks in each BANK are addressed by adopting a logic block;

the algorithm layer constructs RAIDgroup based on logic BANK/Block, thereby realizing physical characteristic independence;

the algorithm module finishes NAND operation by submitting a series of operation requests to the NAND driving module, and each request carries logic addressing information: BANK/logistic Block/Page …

After the NAND driver module receives the request sequence, the following steps are performed as shown in tables 1 and 2; converting the Bank into corresponding DIE/Plane0, and converting the Logical Block into a physical Block;

after Physical addressing information (DIE, Plane, Physical Block, etc.) is acquired, the NAND driver module initiates a corresponding operation to the NAND.

Table 1:

BANK	DIE	Plane
			0	0	0
1	0	1
			2	1	0
3	1	1
			…
2N	N
			0
2N+1	N			1

table 2:

while the invention has been described with reference to a particular embodiment, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. A method for protecting data of a solid state disk comprises the steps of selecting N blocks in each channel to form a transverse RAID stripe group, wherein at least 1 Block is a data check Block, and N is greater than 1; the method is characterized in that the blocks belonging to the same transverse RAID stripe group and belonging to the same channel are at least separated by more than 1 block; respectively establishing logical planes for physical planes except the physical Plane0 in the same channel; carrying out dislocation recombination on the sequence of the block sequence numbers of the logical planes and the sequence of the blocks of the corresponding physical planes, specifically, carrying out recombination through a block logical mapping table, mapping the blocks with the sequence numbers of P1+ M1 in the physical planes to the blocks with the sequence numbers of P1 in the logical planes, if the number of the planes is more than 2, taking different values for M1 selected by different planes, and grouping the blocks of the same channel to the physical Plane0 according to the block sequence numbers of the physical Plane0 when a transverse RAID stripe group is formed; the physical planes other than the physical Plane0 are assigned with the block sequence numbers of the logical planes corresponding thereto.

2. The method according to claim 1, wherein when the solid state disk has more free space, the background scans in blocks when the solid state disk controller is free, and when a data error occurs in a block during data reading, the data is recovered by the horizontal RAID stripe group.