Valentina Timcenko

Cloud Computing in Amazon and Microsoft Azure platforms: Performance and service comparison

This paper addresses the performance and service comparison of two different Cloud Computing platforms: Amazon AWS and Microsoft Azure. Platforms are tested in similar virtual environments, namely for micro instances. Performance is measured by the collection benchmark program called Phoronix Test Suite 3, and the results are presented for Apache and Dbench benchmarks.

Performance examination of type-2 hypervisors: case of particular database application in a virtual environment

This paper presents the results of performance comparison for three different virtualization software, Oracle VM VirtualBox, VMware Player and Microsoft Virtual PC 2007, in Windows 7 operation system environment. All the testing procedures are performed in the same hardware environment and on the same disc partition. We have applied NTFS file system, while Windows 7 is used as host and guest operational systems. Postmark is benchmark application created for the needs of file system performance measurement, and easily used for testing systems and virtualization applications performances based on the different file size.

HbbTV application for DVB-T2

The purpose of this paper is to present an approach to enable the overview of Internet content via digital terrestrial TV. The chosen content will be transmitted over a DVBT/ DVB-T2 broadcast network to the receiver (Set-top-box) and it will be displayed on TV screens. Simple TV application is developed for emulation of Internet content on TV screen, based on HbbTV standard, but in absence of broadband (or any other) Internet connection and without return channel from TV receiver to the repeater.

The comprehensive performance analysis of striped disk array organizations - RAID-0

Redundant Array of Inexpensive Disks version RAID-0 is a block-stripped disk array without redundancy. This paper provides results of RAID-0 performance examination when compared to a single disk drive. Particular emphasis is put on how the number of disks in the RAID 0 configuration influences the performance results. The presented analysis relies on setting up the analytical model of a single disk based on M/G/1 queue and in the same manner defining analytical model for the file system access time. Finally, we have set up an analytical model for the RAID-0 workload related to four specific types of RAID performances. We have defined three types of workloads, generally dominated by relatively small objects. Test results have shown a superior behaviour of RAID-0 when compared to a single drive and a significant performance enhancement as the number of disks increases. All benchmark results are interpreted based on a mathematical model.

Modeling of mobility in urban environment

The paper analyzes the impact of duration and probability of generating a pause in the nodes movement on the performance of AODV routing protocol in an urban environment. Mobility in the urban environment is generated by applying Manhattan Grid mobility model. In addition, we have performed a comparative analysis of results obtained using two mobility generators, BonnMotion and SUMO. Experimental comparison is based on a number of OTcl scripts simulated in NS2 environment and adequate mobility generator. Analysis of the results was performed using Trace Graph application.

Performance comparison of virtualization softwares used on MS Windows 7

This paper presents performance comparison results of three similar virtualization softwares: Oracle VM VirtualBox, VMware Player and Microsoft Virtual PC 2007, under the operating system Windows 7. All tests were performed in the same hardware environment, on the same disk partition. The testing file system is NTFS, while we have used Windows 7 both as a host and as a guest operating system. For performance measurement we have used Postmark benchmark software, while testing procedures were carried out for different file sizes.

The influence of disk number to the performance of striping RAID

This paper presents results related to performance analysis of the number of disk influence to striping RAID configurations under the Linux operating system. The evaluation is based on mathematical modeling of RAID performances. The observed performances were measured by the means of the Postmark benchmark application, taking into cosideration two variations of RAID, striping RAID-0 and RAID-5, under variation of the number of disks employed in RAID striping configuration. The influence of disk performance is interpreted on the basis of a mathematical performance model.

Ensemble classifiers for supervised anomaly based network intrusion detection

This paper focuses on the problem of machine learning classifier choice for network intrusion detection, taking into consideration several ensemble classifiers from the supervised learning category. We have evaluated Bagged trees, AdaBoost, RUSBoost, LogitBoost and GentleBoost algorithms, provided an analysis of the performance of the classifiers and compared their learning capabilities, taking for the reference UNSW-NB15 dataset. The obtained results have indicated that in the defined environment and under analyzed conditions Bagged tree and GentleBoost perform with highest accuracy and ROC values, while RUSBoost has the lowest performances.

Competition of virtualized ext4, xfs and btrfs filesystems under type-2 hypervisor

This paper focuses to the performance analysis of three different 64-bit guest Linux FSs, ext4, xfs and btrfs, under hypervisor type-2 virtualization. Postmark and Bonnie++ benchmarks are used for generation of the testing environment. The emphasis is on the read and writes throughputs measured for different workload conditions. The experiments have indicated that ext4 file system provides the best performances in majority of the tests.

A multi-objective assessment framework for cloud computing

In this paper we propose a multi-objective assessment framework for cloud computing technology. Starting from the technology itself, we perform top down approach, decomposing primary objectives: economy, ecology, business and social, into hierarchical structure. Decomposition is performed until the level at which quantification of objectives is possible and appropriate decision space appears recognizable. To fit the scope of this paper, we provide only representative parts of the framework.