Unlock instant, AI-driven research and patent intelligence for your innovation.

Method and device for predicting burst access behavior of storage system

A technology of storage system and prediction method, which is applied in the field of prediction of sudden access behavior, to achieve the effects of high efficiency, stable frequent association mode, and excellent mining efficiency

Active Publication Date: 2019-03-22
广州联图电子科技有限公司 +1
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, some of the work related to sudden access done by the predecessors mentioned above basically revolves around the phenomenon of sudden access behavior, how to detect sudden access behavior, how to solve the serious consequences of sudden access behavior, and how to adopt There are no studies on predicting bursty behavior, etc.

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 device for predicting burst access behavior of storage system
  • Method and device for predicting burst access behavior of storage system
  • Method and device for predicting burst access behavior of storage system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] Such as figure 1 As shown, the storage system-oriented prediction device for burst access behavior in this embodiment includes an association mining module, an association filtering module, and a burst prediction module, wherein:

[0050] The associated mining module is used to take a part of the I / O (Input / Output, input / output) data set of the storage system as a training set, and carry out frequent associated I / O mining to the training set;

[0051] The association filtering module is used to divide the test set of the I / O data set of the storage system with the specified forecast time as the time granularity, and obtain the total number of I / O requests at the corresponding forecast time point of the test set; then extract each Predict the I / O request data in the specified observation time at the time point, filter the I / O request data at each observation time point, filter the frequently associated I / O at each observation time point, and keep each observation time P...

Embodiment 2

[0073] This embodiment is a specific application example of the method for predicting burst access behavior of a storage system in Embodiment 1.

[0074] to combine Figure 3a~3c Workflow diagram of the associated hardening window matching and Figure 4 A detailed analysis of the workflow flow chart of sudden access behavior prediction.

[0075] In order to make the Apriori algorithm more advantageous in mining such data sets used in this embodiment, a kind of Apriori algorithm based on the correlation strengthening window is proposed. The overall idea of ​​the algorithm follows the idea of ​​the Apriori algorithm, but in the processing In the third step, when scanning the database to match the candidate set with the entire data set, counting the support of the candidate set and deleting the candidate sets that do not meet the requirements, a correlation strengthening window is introduced, which is used to limit the data block. The amount of data, because each data block cor...

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 prediction method and device for burst access behaviors oriented to a storage system. The method comprises following steps: fetching parts of I / O datasets in the memory system as training sets and digging frequent association I / O of the training sets; dividing test sets of the I / O dataset in the memory system by utilizing specified predication time as time granularity and acquiring the total number of I / O requests for prediction time points corresponding to the test sets; extracting I / O request data in specified observation time at each predication time point and filtering I / O request data at each observation time point such that frequent association I / O is filtered at each observation time point and non-frequent association I / O at each observation time point is preserved; and employing non-frequent association I / O at each observation time point to judge whether burst access behaviors occur or not at corresponding prediction time points. The prediction method and device for burst access behaviors in the memory system is capable of adjusting training sets of I / O datasets in the storage system in a flexible manner and acquired frequent association I / O is used for a subsequent burst predication experiment.

Description

technical field [0001] The invention relates to a prediction method and device for sudden access behavior, in particular to a storage system-oriented prediction method and device for sudden access behavior, belonging to the fields of data mining and data storage. Background technique [0002] Burst access is a very common data access pattern, which includes various computer-related workloads (such as disk I / O, file system access, network packets, web page access, etc.). Burst refers to the formation of a large number of workloads in a very short period of time, causing the server to fail to meet the sharply rising traffic and to successfully complete related requests submitted by users. For example, in the 2008 Olympic Games, due to large-scale access requests, the booking system collapsed and could not be accessed smoothly. Another example is the 1998 World Cup when data access suddenly surged to a very high level during the live broadcast, and then returned to normal acce...

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): G06F9/50
Inventor 邓玉辉黄立锋
Owner 广州联图电子科技有限公司