Presentation on theme: "An Overview of ABFT in cloud computing Shiv Kumar Gupta Department of Computer Science A.P.S.U., Rewa M.Phil.(2009-10)"— Presentation transcript:
An Overview of ABFT in cloud computing Shiv Kumar Gupta Department of Computer Science A.P.S.U., Rewa M.Phil.(2009-10)
An Overview of ABFT in cloud computing Introduction Dependability Fault-tolerance ABFT FFT Conclusions
What is cloud computing? Cloud computing gets as a metaphor for the internet. The internet is represented in network diagram as a cloud. A cloud is a type of parallel and distributed system. It is a collection of inter-connected and virtualized computers
What is cloud computing? Which are presented as one or more unified computing resources. It is based on service – level agreements. Its communication between service provider and consumer. So, I can say in the internet of search engine like www.google.com is cloud viz.,www.google.com
. G – Global O – Outsourcing O – Of G – Grounded L – Learning E – Encyclopedia/Era Is cloud computing over the internet
Meaning of dependability Dependability is the concept of reliability,safety or security. It is tool of attributes, means, impairments Attributes have availability, reliability, safety, confidentiality, integrity, and Maintainability. Means have fault prevention, fault tolerance, fault removal, fault forecasting Impairments have faults, errors, failures
Fault -tolerance Fault tolerance methods come into play the moment a fault enters the system boundaries. Fault tolerance have two sub-techniques error processing and fault treatment. Error processing aims at removing errors from the computational state. Computational state means if possible, before failure occurrence.
Fault-tolerance In the error processing technique task is error detection, error diagnosis, and error recovery. And fault treatment aims at preventing faults from being re-activated.
ABFT Algorithm-based fault tolerance technique presented by Hung and Abraham in 1984 ABFT is aimed at hardening processors when executing matrix application. Hardening processor is obtained by adding,coding information to matrices. Matrices application is working on n x m elements.
ABFT Corresponding column check sum matrix A c is an (n+1) x m matrix. Which consists of the matrix A in the first n rows and a column summation vectors in the (n+1)th row. A Checksum Figure 1-1. A column
ABFT Corresponding row checksum matrix A, is an n x (m+1) matrix. Which consists of the matrix A in the first m column and a row summation vectors in the (m+1)th column. A checksum Figure1-2. A row
ABFT Corresponding full checksum matrix A f is an (n+1) x (m+1) matrix. Which is the column checksum matrix of the row checksum matrix A r of A. A Checksum checksum Figure 1-3. A Checksum Matrix
FFT Fast Fourier Transformation is another Algorithm based technique. Which is an ad hoc hardware architecture. Supported to ad hoc hardware. The main purpose to modify the algorithm to detect faults arising in the ad hoc hardware architecture.
FFT FFT method is suitable in conventional computer. Even if the FFT algorithm is implemented in software on a conventional computer
Conclusion Algorithm based approach provides good results in terms of detection capability and relatively. And also in terms of memory and performance overhead. Limitation of algorithm based approach has been only applied to those algorithm for which an ABFT created. Mainly to regular data structure.