Fiaz Gul Khan

Performance evaluation of fault tolerance techniques in grid computing system

As fault tolerance is the ability of a system to perform its function correctly even in the presence of faults. Therefore, different fault tolerance techniques (FTTs) are critical for improving the efficient utilization of expensive resources in high performance grid computing systems, and an important component of grid workflow management system. This paper presents a performance evaluation of most commonly used FTTs in grid computing system. In this study, we considered different system centric parameters, such as throughput, turnaround time, waiting time and network delay for the evaluation of these FTTs. For comprehensive evaluation we setup various conditions in which we vary the average percentage of faults in a system, along with different workloads in order to find out the behavior of FTTs under these conditions. The empirical evaluation shows that the workflow level alternative task techniques have performance priority on task level checkpointing techniques. This comparative study will help to grid computing researchers in order to understand the behavior and performance of different FTTs in detail.

A hybrid fault tolerance technique in grid computing system

In order to achieve high level of reliability and availability, the grid infrastructure should be a foolproof fault tolerant. Fault tolerance plays a key role in order to assert availability and reliability of a grid system. Since the failure of resources affects job execution fatally, fault tolerance service is essential to satisfy QoS requirement in grid computing.In this paper we proposed two hybrid fault tolerance techniques (FTTs) that are called alternate task with checkpoint and alternate task with retry. These proposed hybrid FTTs inherit the good features and overcome the limitations of workflow level FTT and task level FTT. We evaluate the performance of our proposed FTTs under different experimental environments. Finally, we conclude that alternate task with checkpoint improves the reliability of a grid system more significantly than alternate task with retry.

Analysis of Fast Parallel Sorting Algorithms for GPU Architectures

Sorting algorithms have been studied extensively since past three decades. Their uses are found in many applications including real-time systems, operating systems, and discrete event simulations. In most cases, the efficiency of an application itself depends on usage of a sorting algorithm. Lately, the usage of graphic cards for general purpose computing has again revisited sorting algorithms. In this paper we extended our previous work regarding parallel sorting algorithms on GPU, and are presenting an analysis of parallel and sequential bitonic, odd-even and rank-sort algorithms on different GPU and CPU architectures. Their performance for various queue sizes is measured with respect to sorting time and rate and also the speed up of bitonic sort over odd-even sorting algorithms is shown on different GPUs and CPU. The algorithms have been written to exploit task parallelism model as available on multi-core GPUs using the OpenCL specification. Our findings report minimum of 19x speed-up of bitonic sort against odd-even sorting technique for small queue sizes on CPU and maximum of 2300x speed-up for very large queue sizes on Nvidia Quadro 6000 GPU architecture.

Aggregated provenance and its implications in clouds

Abstract Cloud computing follows a layered architecture where each layer targets a particular domain of end users. In such a layered architecture, provenance (the metadata that describes the derivation history of the object) of the individual layers is of significant importance to establish trust and authenticity. In a typical Cloud environment, each layer provides important provenance information which usually targets a particular domain of clients e.g. Cloud provider uses infrastructure provenance to track resource utilization. In the case of aggregated provenance, it becomes challenging to manage provenance information because of the relationships that exist within Cloud layers and the creator object. The existing techniques and systems to address provenance in Clouds usually work at a single layer of abstraction. These systems, however, fail to answer questions which require aggregated provenance from the individual layers. In this paper, we reason about the need of aggregated provenance, its significance and a proposed solution which works at different layers of abstraction for the management of provenance data. We also present a case study to signify the importance and necessity of collective provenance for various application domains.

A Mutual Demagnetizing Tensor for N-Body Magnetic Field Modeling

We introduce a mutual demagnetizing tensor for calculating the demagnetizing field in multiple magnetic bodies. Algorithms for magnetic simulations dealing with interactive n-bodies usually treat the simulation domain as a single magnetized body with embedded nonmagnetic regions. Doing so minimizes the number of variables used in the field calculations but the computation time and storage requirements can become very large. Our approach reduces memory consumption and shows a gain in performance for cases involving field calculation in patterned films. We have verified our results by performing simulation of dots in periodic grids and report a speed-up ranging between 3 × and 10 ×.

Computation of Demagnetization Tensors by Utilizing Fourier Properties

We describe a method for the computation of the discrete demagnetization tensor on regular cuboid grids. Assuming homogeneously magnetized cells, the tensor components can be calculated exactly by known analytical formulas. These integral-based expressions can be expensive due to their nested nature and difficult to code. The novelty of this paper is that parts of the tensor computation are moved to Fourier space, which simplifies the implementation. The main idea is that some nested sums, required for the computation of the tensor in real space, are replaced by simple multiplications with real factors in Fourier space. For regular grids, the demagnetization field is usually computed in Fourier space by application of the convolution theorem. Thus, computing the tensor in Fourier space in the first place does not introduce any drawbacks.

Introducing ToPe-FFT: An OpenCL-based FFT library targeting GPUs

In this paper, we present our implementation of the fast Fourier transforms on graphic processing unit (GPU) using OpenCL. This implementation of the FFT (ToPe‐FFT) is based on the Cooley‐Tukey set of algorithms with support for 1D and higher dimensional transforms using different radices. Factorization for mix‐radices enables our code to target FFTs of near arbitrary length. In systems with multiple graphic cards (GPUs), the library automatically balances the FFT computation thus achieving maximum resource utilization and higher speedup. Based on profiling and micro‐benchmarking of ToPe‐FFT, it is observed that the average speedup of our library for different sizes is 48× faster than the single CPU‐based code using FFTW and 3× faster than NVIDIAs GPU‐based cuFFT library.

Review of parallel and distributed architectures for micromagnetic codes

Purpose – Rapid advancements in computer technologies over the past decade have recorded significant growth in the area of computational micromagnetics. As a result, current micromagnetic codes exploit the scalability offered by parallel and distributed computer architectures to deliver maximum performance. The purpose of this paper is to present a review, which explores various aspects of this relationship. Design/methodology/approach – The authors arrange the theme for this paper around the micromagnetic code development process. The review involves a discussion of the micromagnetic model, some new parallel architectures, and computational aspects based on different numerical methods. Findings – As current micromagnetic code is not readily portable to different architectures, most of the development effort goes towards this area, with a focus on writing/rewriting code for streaming hardware (particularly graphic cards). Originality/value – The paper identifies key challenges and avenues for further rese...

Parallel butterfly sorting algorithm on GPU

Efficient sorting is vital for overall performance of the underlying application. This paper presents Butterfly Network Sort (BNS) for sorting large data sets. A minimal version of the algorithm Min-Max Butterfly is also shown for searching minimum and maximum values in data. Both algorithms are implemented on GPUs using OpenCL exploiting data parallelism model. Results obtained on different GPU architectures show better performance of butterfly sorting in terms of sorting time and rate. The comparison of butterfly sorting with other algorithms:bitonic, odd-even and rank sort show significant speedup improvements against all on Nvidia Quadro-6000 GPU with relatively better sorting time and rate

Secure-CamFlow: A device-oriented security model to assist information flow control systems in cloud environments for IoTs

Recent developments in the cloud technologies have motivated the migration of distributed large systems, specifically the Internet of Things to the cloud architecture. Since Internet of Things consist of a vast network and variety of objects, the cloud platform proves to be an ideal option. It is essential for the proper functioning of the Internet of Things to be able to share data among the system processes. The biggest problem faced during the transition of the IoTs to the cloud is the security of data especially while data sharing within the cloud and among its tenants. Information Flow Control mechanisms are one of the many solutions to enable a controlled sharing of data. Integration of Information Flow Control Systems to the existing architecture requires various levels of re‐engineering efforts. Moreover, most of the Information Flow Control systems focus on data flow within the cloud and neglect the security and integrity of data while it is being transferred to the cloud from various devices. This research focuses on securing the entire process of data migration to cloud from devices while the in‐cloud data flow is monitored by the Information Flow Control policies specified by the users. We have developed a prototype for the proposed model, and results are evaluated on the basis of energy consumption and execution time. As proposed model provides security services such as privacy, integrity, and authentication, hence it takes more execution time and consumes more energy as compared with the existing model.