A Survey of Software Fault Tolerance Techniques Authors: Zaipeng - - PDF document

1

A Survey of Software Fault Tolerance Techniques

Authors: Zaipeng Xie, Hongyu Sun and Kewal Saluja Dependable Software Design By: M. Abdollahi Azgomi

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 2

Outline

Introduction Types of recovery

Backward recovery Forward recovery

Traditional Software Fault Tolerance

Design diversity Recovery blocks (RcB) N-version programming (NVP) N-self checking programming (NSCP) Data diversity N-copy programming (NCP) Retry blocks Environmental diversity

2

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 3

Introduction

What Is Software Fault Tolerance?

Provide service complying

)ﻦﻴﻣﺎﺗ ،ندروآﺮﺑ( with the specification in spite of software faults

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 4

Strategies of Software Fault Tolerance

Error Processing

Error Compensation

)ﺎﻄﺧ ناﺮﺒﺟ(

Error Recovery

)ﺎﻄﺧ ﺖﻓﺎﻳزﺎﺑ(

Forward Error Recovery (FER) Backward Error Recovery (BER)

Fault Treatment

) ﺎﻄﺧ نﺎﻣرد(

3

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 5

Typical Software FT Techniques

Single version software FT techniques (mostly in self-

checking software)

Multiple version software FT techniques (design

diversity)

recovery blocks (RcB), N-version programming (NVP) and N self-checking programming (NSCP).

Multiple data representation techniques (data diversity)

retry blocks (RtB) and N-copy programming (NCP)

Temporal diversity

)ﻲﻧﺎﻣز عﻮﻨﺗ(

Environment diversity

)ﻲﻄﻴﺤﻣ عﻮﻨﺗ(

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 6

Types of Recovery

Error recovery is performed using

backward recovery

)ﺐﻘﻋ ﻪﺑ ور ﺖﻓﺎﻳزﺎﺑ(

• r

forward recovery

)ﻮﻠﺟ ﻪﺑ ور ﺖﻓﺎﻳزﺎﺑ(.

4

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 7

Backward Recovery

Attempts to return the system to a correct or error-

free state.

For transient faults Example:

recovery blocks (RcB)

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 8

Forward Recovery

Attempts to find a new state from which the

system can continue operation.

Utilize error compensation based on redundancy

to select or derive the correct answer or an acceptable answer.

Examples:

N-version programming (NVP ),

)ﻲﺷرﺎﮕﻧﺪﻨﭼ(

N-copy programming (NCP) and

)يا ﻪﺨﺴﻧﺪﻨﭼ(

the distributed recovery block (DRB)

5

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 9

Forward Recovery

Efficient For Predictable

Errors

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 10

Traditional Software Fault Tolerance

• Software design and implementation errors cannot be

detected by simple replication of identical software units, assuming the same inputs are provided to each copy.

• Some form of diversity must accompany the

redundancy

• Software redundancy Design diversity
• Information or data redundancy Data diversity
• Temporal redundancy Temporal diversity
• Environment diversity
• Hardware redundancy

6

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 11

Design Diversity

Higher cost

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 12

Software Fault Tolerance Techniques Using Design Diversity

Error detection by AT and forward recovery NSCP N Self-Checking Programming Vote NVP N-Version Programming Error detection by AT and backward recovery RcB Recovery Blocks Error Processing Abbr. Techniques

7

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 13

RcB

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 14

Recovery Blocks

Sequential Execution Scheme Absolute, with respect to specification Criteria of Accepting Result Error detection by AT and backward recovery Error Processing Technique Recovery block Method

8

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 15

NVP

• DM: Decision

Mechanism

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 16

N-version Programming

Parallel (on NMR hardware)

Execution Scheme

Relative, on variant results

Criteria of Accepting Result

Vote

Error Processing Technique

N-version programming

Method

9

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 17

NSCP

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 18

N Self-Checking Programming

Parallel

Execution Scheme

Relative, on variant results or Absolute with respect to specification

Criteria of Accepting Result

Error detection and result switching Then, Detection by comparison or by AT(s)

Error Processing Technique

N self-checking programming

Method

10

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 19

Data Diversity

To complement design diversity Using data re-expression algorithms (DRA) to obtain

logically equivalent variants of the input data

Data re-expression via decomposition and recombination

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 20

Software Fault Tolerance Techniques Using Data Diversity

Run the same process concurrently or sequentially NCP N-Copy Programming Acceptance test and Backward recovery RtB Retry Blocks Error Processing Abbr. Software Fault Tolerance Techniques

11

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 21

Retry Blocks

Error Pass

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 22

Retry Blocks

Sequential Execution Scheme Absolute, with respect to specification Criteria of Accepting Result Error detection by AT and backward recovery by DRA Error Processing Technique Retry blocks Method

12

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 23

NCP

• Inputs are

re-expressed

DSD - A Survey on Sowftare FT Techniques - By. M. A. Azgomi - IUST-CE 24

N-copy Programming

Parallel Execution Scheme Relative, on variant results Criteria of Accepting Result Decision mechanism (DM) and forward recovery Error Processing Technique N-copy programming Method