Dynamic Fault-Tolerant Elastic Scheduling Method

A scheduling method and elastic technology, applied in the field of cloud computing, can solve problems such as increased failure probability, disk drive damage, 4%, etc., to achieve the effect of balance, high performance, reliability and resource optimization

Active Publication Date: 2019-04-05
NAT UNIV OF DEFENSE TECH
View PDF9 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Additionally, about 5% of disk drives fail, and servers crash at least twice a year, a failure rate of 4%
To make matters worse, in order to reduce production costs, data centers often use less expensive commodity hardware, resulting in an increased probability of failure in the cloud

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
  • Dynamic Fault-Tolerant Elastic Scheduling Method
  • Dynamic Fault-Tolerant Elastic Scheduling Method
  • Dynamic Fault-Tolerant Elastic Scheduling Method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0079] Example 1: when the task in the VM When the above is weak fault tolerance.

[0080] In this case, by the definition of weak fault tolerance, the task It is quite possible to meet the deadline. To measure the likelihood of completion, the Defined as completing a task on a VM The probability:

[0081]

[0082] For clouds, there may be multiple virtual machines belonging to Weak fault tolerance, therefore, The global optimal scheduling strategy is to find each of The maximum value of , defined as , however, for lower , if it cannot be completed before the deadline, it will not only waste resources, but also affect the execution of subsequent tasks. Therefore, it is necessary to determine a threshold As a metric for determining scheduling policies.

[0083] Further, the decision tree method can be used to determine the threshold ,Such as Figure 5 As shown in , it can be seen that the decision tree has three outputs, "Completed", "Rejected" an...

Embodiment 2

[0089] Example 2: when the task in the VM When the above is strong fault tolerance.

[0090] In this case, the task It is always possible to meet the deadline. Considering overall resource optimization, the key issue is to determine when to use the replication mechanism and when to use the central commit mechanism.

[0091] when set up, initial task and the resubmitted task There will be a time overlap between them. However, if resubmission is used, the original task was in When it fails later, the resubmitted task does not have enough time slots to complete before the deadline. In this case, resubmitting cannot achieve effective fault tolerance. Therefore, to achieve effective fault tolerance, a replication mechanism must be employed.

[0092] if holds, there may be no time overlap, so the resubmission mechanism is available. but if just than Slightly larger, it may not have enough time for resubmission tasks to execute, such as Figure 6 shown. In t...

Embodiment 3

[0094] Embodiment 3: Using task scheduling analysis when resubmitting.

[0095] Due to the fluctuation of the running time of the task, the resubmission mode will have many forms, and these forms are abstracted into two categories.

[0096] Category 1:

[0097] Such as Figure 7 shown, where the arrow indicates the ready time . assumed can be satisfied.

[0098] Therefore, if , regardless of the initial task Whether or not it completes successfully, there is always enough time slot to ensure completion of the resubmitted task. According to the basic assumption of the fault model, if fail, can be completed successfully. Therefore, effective fault tolerance can be achieved.

[0099] Category 2:

[0100] Such as Figure 8 shown, if exist failure, where ,So should start from Begin execution. However, if the actual task runtime , the system cannot be fault-tolerant in this case. Therefore, the effective fault tolerance probability using the resu...

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

A dynamic fault-tolerant elastic scheduling method comprises the following steps: estimating a task running time; determining a scheduling type of the task according to a flexible time situation of the task; according to the scheduling type of the task, selecting the virtual machine executing the task from the virtual machine list and scheduling to realize the reliability and resource optimizationbetween the clouds.

Description

technical field [0001] The invention relates to the technical field of cloud computing, in particular to a dynamic fault-tolerant elastic scheduling method. Background technique [0002] Cloud computing has become a mainstream distributed computing model for on-demand configuration of computing resources to meet dynamic application workloads. Running applications on virtual resources, specifically virtual machines (VMs), is an efficient solution that improves economics and scalability. In practice, many applications face the problem of ever-increasing data growth and high computational complexity, such as astronomy, financial transactions, and physics. Cloud computing can effectively support the high computing performance requirements of these scientific applications. There are reports that for a cloud of 10,000 exceptionally reliable servers (with a mean time between failures of 30 years), there will be one failure per day. Additionally, about 5% of disk drives fail, and...

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): G06F9/455G06F9/50G06F11/14
CPCG06F9/45558G06F9/5038G06F11/1461G06F2009/45562
Inventor 朱晓敏闫辉包卫东王吉陈黄科吴冠霖陈俊杰张国良张雄涛张亮
Owner NAT UNIV OF DEFENSE TECH
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