CN107545068A

CN107545068A - The data of the memory of fixed size forget and recovery system

Info

Publication number: CN107545068A
Application number: CN201710837669.XA
Authority: CN
Inventors: 陈秋朴
Original assignee: Individual
Current assignee: Individual
Priority date: 2017-09-18
Filing date: 2017-09-18
Publication date: 2018-01-05

Abstract

The memory near infinite storage system of fixed size, it is characterized in that the system includes documents in my module and file access pattern module, wherein documents in my module be used for it is hardly important to memory and for a long time without using a data part forget, file access pattern module is used to recover the data of forgetting.

Description

The data of the memory of fixed size forget and recovery system

Technical field

The present invention and storage device are related to some compress techniques and data recovery technique.

Background technology

With the rise of machine learning, artificial intelligence rapid development, but artificial intelligence is very big to the demand of data, and this is right The capacity requirement of memory is very high, although cloud can solve some problems, in the area that no network, network are shielded Domain or when can not network, when memory space takes full, machine can not just be run artificial intelligence robot, just lose artificial intelligence The value of energy.

The content of the invention

It is an object of the invention to provide a kind of forgetting of the data of memory of fixed size and recovery system, realize and be not required to A part for hardly important data can be forgotten as the mankind in the case of expanding memory space, and need Data memory can be carried out when wanting, so as near infinite data storage, substantially increase memory utilization rate and The speed of machine learning.

The technical solution used in the present invention is：

Make file close to the data and some supplementary numbers of complete deletion state, only reservation core using documents in my technology According to greatly reducing the size of file, release substantial amounts of memory space；Make file access pattern using file access pattern module, pass through Some core datas and assistance data of remaining remove restoring data as far as possible；

The present invention the characteristics of：The space of storage is substantially reduced using compress technique；Using File Instauration Technique, make extensive The accuracy of file after multiple greatly improves；Using Multiple Compression technology, make the size of file very small, close to nothing, and Recover to be able to preserve due to core data and make the accuracy of recovery data very high by the association of assistance data during file.

Brief description of the drawings

Fig. 1 is the module that the data of the memory of the fixed size of the present invention are forgotten and recovery system includes.

Fig. 2 is the flow chart of the forgetting module of the file of the present invention.

Fig. 3 is the flow chart of the recovery module of the file of the present invention.

Embodiment

The present invention is described in detail with reference to the accompanying drawings and detailed description.

The data forgetting of the memory of fixed size of the present invention includes modular structure with recovery system, as shown in figure 1, bag Module containing documents in my and file access pattern module.

With reference to shown in Fig. 2 program flow diagrams, the forgetting module of file realizes that step is as follows：

The data of the memory of fixed size forget and recovery system not timing check external memory whether soon by Take 2.When memory is soon occupied full, this system checks whether current hosts are in busy condition 3.

Further, if current hosts are not at busy condition, file 4 remote and hardly important is replicated, so After check this file format；Otherwise the 2nd step is returned.

Further, if file format be those significant datas the centre of data (such as：Picture etc.) 5, replicating Four weekly datas of file select a fritter at random data reset in block, plus the big of flag bit and randomized block in randomized block It is small, and compress the file 6 of this duplication；If file format is that those data distributions are uniform (such as：Audio, video, document etc.), A fritter is selected at random data reset in block in the data of the file of duplication, plus flag bit and randomized block in randomized block Size, and compress the file 7 of this duplication.

Further, compressed file afterbody successively plus original initial address, replicate file initial address simultaneously Start mark position is added before initial address, original is deleted and replicates file 8, terminate to forget file operation.

If necessary, file can use forgetting technology multiple, you can be compressed by reset and add original text each time in afterbody The initial address of the memory of part is multiple.

With reference to shown in Fig. 3 program flow diagrams, the recovery module of file realizes that step is as follows：

The data of the memory of fixed size are forgotten and recovery system finds the forgetting file to be recovered, and find forgetting file Afterbody, and judge whether tail of file has start mark position and initial address 9.

Further, if tail of file has start mark position and initial address, the size identical with original is found Space, the corresponding data decompression for replicating this file is copied into this space 10；Otherwise, terminate to recover file operation.

Further, file 11 is found according to initial address.

Further, judge whether this file is destroyed 12.

Further, if this file is destroyed, return to whether tail of file has the behaviour of start mark position and initial address 9 Make.

Further, if this file is not destroyed, a compression section of this file, which copies to, forgets putting for file 13 in the randomized block of " 0 ".

Further, the data whether correct 14 of the file in space are judged.

Further, it is if correct right, terminate to recover file operation；Otherwise, by analyzing available data, associating progress Data conjecture makes file close to correct, end recovery file operation.

Claims

1. a kind of data of memory of fixed size forget and recovery system, it is characterised in that the system includes documents in my mould Block (1) and file access pattern module (2)；Documents in my module (1) be used for it is hardly important to memory and for a long time without using number According to a part forgotten；File access pattern module (2) is used to recover the data of forgetting.

2. according to the storage system described in claim 1, it is characterised in that documents in my module (1) includes checking that memory is It is no to be occupied full part soon, judge the whether busy part of computer, duplication remote and hardly important file part, check text Whether part form is significant data in the core of data, randomized block reset and compressed file part, deletion original and answers These parts of file part processed are formed.

3. according to the storage system described in claim 1, it is characterised in that file access pattern module (2) is by judgement tail of file It is no to have start mark position and initial address part, replicate and forget file part, file part is found according to initial address, judges text The whether destroyed part of part, the whether correct part of the data for copying data to randomized block part, judging the file after recovering and connection Think that data guess part.

4. according to the storage system described in claim 2, it is characterised in that simultaneously compressed file part includes weight for randomized block reset Want data the centre of file compressing file and significant data the surrounding of file compressing file.

5. according to the storage system described in claim 2, it is characterised in that delete original and duplication file part is included in pressure Contracting tail of file adds the initial address of original, replicates the initial address of file and before initial address plus starting mark successively Remember position, delete original and replicate file.

6. according to the storage system described in claim 3, it is characterised in that duplication forgetting file part includes finding and original Size identical space, replicate this file corresponding data decompression be copied into this space.

7. according to the storage system described in claim 4, it is characterised in that compressing file bag of the significant data in the centre of file Four weekly datas included in the file of duplication select a fritter data reset in block at random, in randomized block plus flag bit and with The size of machine block, and compress the file of this duplication.

8. according to the storage system described in claim 4, it is characterised in that compressing file bag of the significant data in the surrounding of file The data included in the file of duplication select a fritter that data reset in block is added flag bit and randomized block in randomized block at random Size, and compress this duplication file.