Method and system for hybrid memory oriented power failure recovery

A hybrid memory and power-off recovery technology, applied in the field of big data, can solve problems such as difficult practical application, high cost of non-volatile memory, and MemKV technology development limitations, and achieve high-performance, high-reliability power-off recovery effects

Active Publication Date: 2020-11-27
PEKING UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Non-volatile memory is expensive and difficult to implement
Therefore, the existing MemKV persistence technology is difficult to simultaneously support high performance (able to quickly complete read and write operations) and high reliability (able to quickly and completely complete power-off recovery), which limits the development of MemKV technology

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Method and system for hybrid memory oriented power failure recovery
  • Method and system for hybrid memory oriented power failure recovery
  • Method and system for hybrid memory oriented power failure recovery

Examples

Experimental program
Comparison scheme
Effect test

Embodiment example 1

[0049] In this implementation example, the index is organized in a B+ tree structure, all leaf nodes and non-leaf nodes are in order, and the numbers and addresses of all files are stored in the leaf nodes. The default recovery time node of the system is the power-off time. The default recovery level of the system is the first recovery level.

[0050] The recovery process is as follows:

[0051] (1) The system loses all content in DRAM due to abnormal power failure.

[0052] (2) After the power is turned on, the user starts the machine, and the operating system is automatically loaded.

[0053] (3) Since the recovery time node is the power-off time, all entries in the metadata area are retained.

[0054] (4) sequentially read 13 pieces of metadata in the metadata area, and rebuild the index area according to the generation rules of the B+ tree index (common knowledge in this field), the index area after recovery is as follows Figure 5 shown.

[0055] (5) Restore the hot ...

Embodiment example 2

[0068] In this implementation example, the index is organized in a B+ tree structure, all leaf nodes and non-leaf nodes are in order, and the numbers and addresses of all files are stored in the leaf nodes. The recovery time node specified by the user is December 10, 2018. The user-specified recovery level is the second recovery level.

[0069] The recovery process is as follows:

[0070] (1) The user actively triggers system recovery, and the system restarts immediately.

[0071] (2) Immediately, all of the DRAM is lost, and the operating system is automatically loaded.

[0072] (3) Items 1-6 in the metadata area are retained, items 7-13 are deleted, and entries 7-13 in the data area are deleted.

[0073] (4) Sequentially read the remaining 6 pieces of metadata in the metadata area, and rebuild the index area according to the generation rules of the B+ tree index (common knowledge in this field), and the restored index area is as follows: Figure 7 shown.

[0074] (5) Re...

Embodiment example 3

[0079] In this implementation example, the index is organized using a hash structure, and the numbers and addresses of all files are stored in the specified hash bucket. The hash algorithm is to find the remainder after dividing by 7. The default recovery time node of the system is the power-off time. The default recovery level of the system is the third recovery level.

[0080] The recovery process is as follows:

[0081] (1) The system loses all content in DRAM due to abnormal power failure.

[0082] (2) After the power is turned on, the user starts the machine, and the operating system is automatically loaded.

[0083] (3) Since the recovery time node is the power-off time, all entries in the metadata area are retained.

[0084] (4) sequentially read 13 pieces of metadata in the metadata area, and rebuild the index area according to the generation rules of the hash index (common knowledge in this field) and the hash algorithm specified by the user. The restored index are...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a power failure recovery system and a power failure recovery method for a hybrid memory. The method comprises: data and indexes of a MemKV storage system are stored in the hybrid memory formed by an NVM and a DRAM, when the power failure occurs, the indexes are lost, but the data are not lost, and the indexes are recovered through the data, so that the power failure recovery is realized; wherein the hybrid memory comprises the following modules: an operation system area, a user area, an index area, a hot data area, a metadata area and a data area; wherein the operatingsystem area, the user area, the index area and the hot data area are stored in the DRAM; the metadata area and the data area are stored in an NVM. By the adoption of the technical scheme, the MemKV storage system can complete read-write operation rapidly and complete power failure recovery rapidly and completely at the same time, and high performance and high reliability can be met at the same time.

Description

technical field [0001] The invention relates to the technical field of big data, in particular to a hybrid memory-oriented power-off recovery method and system. Background technique [0002] With the advent of the era of big data, the memory-based key-value storage system (MemKV for short) represented by Redis and Aerospike has ushered in rapid development. In order to obtain extremely high read and write performance, MemKV stores all data and indexes in the memory, but because the traditional DRAM (Dynamic Random Access Memory, dynamic random access memory) memory is volatile after power failure, the reliability of MemKV is difficult to guarantee , Whether it is a power failure caused by active shutdown or an accidental disaster power failure (hereinafter collectively referred to as power failure), all data and indexes in the DRAM memory will be lost. [0003] In order to achieve a balance between high performance and high reliability, there are two main types of existing ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06F11/14
CPCG06F11/1435G06F11/1441
Inventor 王平杨洪章
Owner PEKING UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap