Vojislav B. Misic

Performance Analysis of Cloud Computing Centers Using M/G/m/m+r Queuing Systems

Successful development of cloud computing paradigm necessitates accurate performance evaluation of cloud data centers. As exact modeling of cloud centers is not feasible due to the nature of cloud centers and diversity of user requests, we describe a novel approximate analytical model for performance evaluation of cloud server farms and solve it to obtain accurate estimation of the complete probability distribution of the request response time and other important performance indicators. The model allows cloud operators to determine the relationship between the number of servers and input buffer size, on one side, and the performance indicators such as mean number of tasks in the system, blocking probability, and probability that a task will obtain immediate service, on the other.

Performance of a beacon enabled IEEE 802.15.4 cluster with downlink and uplink traffic

The performance of an IEEE 802.15.4 compliant network operating in the beacon enabled mode with both downlink and uplink traffic is analyzed through discrete time Markov chains and the theory of M/G/1 queues. The model considers acknowledged transmissions and includes the impact of different network and traffic parameters such as the packet arrival rate, packet size, inactive period between the beacons, and the number of stations. We investigate the nonsaturation region and outline the conditions under which the network abruptly goes to saturation. The analysis of stability of the network queues shows that the stability of the downlink queue at the coordinator is the most critical for network operation. Due to the abruptness with which the switch from nonsaturation to saturation occurs, the network operating point has to be carefully chosen according to the volume of downlink traffic. Furthermore, our model shows that certain features prescribed by the standard actually limit the performance of 802.15.4 networks.

The impact of MAC parameters on the performance of 802.15.4 PAN

The operation of a personal area network, operating under the IEEE Standard 802.15.4 in the beacon enabled mode, is analyzed using the theory of discrete time Markov chains and M/G/1/K queues. The model includes the impact of different parameters such as packet arrival rate, number of stations, stations buffer size, packet size, and inactive period between the beacons. We have derived several important performance parameters such as probability of access, probability that medium is idle, queue length distribution in the device, and probability distribution of the packet service time. Some important conclusions regarding the implementation of 802.15.4 networks and compatible network devices are outlined.

Performance of IEEE 802.15.4 beacon enabled PAN with uplink transmissions in non-saturation mode - access delay for finite buffers

In this work, we derive the probability distribution of access delay and calculate throughput of a personal area network operating under the IEEE standard 802.15.4 in the beacon enabled mode. We model the network using the theory of discrete time Markov chains and M/G/1/K queues. The model considers acknowledged uplink transmission in non-saturation mode, and includes the impact of different parameters such as packet arrival rate, number of stations, the finite size of individual node buffers, packet size, and inactive period between the beacons. Our model also captures the problem of congestion at the beginning of the superframe due to multiple transmissions delayed from the previous superframe. The results show that average access delays, even for small buffer sizes, may be quite high if the throughput exceeds 50%, which can seriously affect applications with delay bounds. Values of throughput larger than 50% can be achieved at the expense of larger buffer sizes, which imposes implementation problems on devices with small memory resources.

A Fine-Grained Performance Model of Cloud Computing Centers

Accurate performance evaluation of cloud computing resources is a necessary prerequisite for ensuring that quality of service parameters remain within agreed limits. In this paper, we employ both the analytical and simulation modeling to addresses the complexity of cloud computing systems. Analytical model is comprised of distinct functional submodels, the results of which are combined in an iterative manner to obtain the solution with required accuracy. Our models incorporate the important features of cloud centers such as batch arrival of user requests, resource virtualization, and realistic servicing steps, to obtain important performance metrics such as task blocking probability and total waiting time incurred on user requests. Also, our results reveal important insights for capacity planning to control delay of servicing users requests.

Modeling Bluetooth piconet performance

The performance of a single Bluetooth piconet is analyzed using the theory of M/G/1 queues with vacations. Analytical results for probability distributions of packet access time and service cycle time are derived. Two scheduling policies are modeled and compared: exhaustive service was found to perform better, but limited service does not incur the risk of starvation. A hybrid scheme known as k-limited scheduling is shown to provide a reasonable tradeoff between performance and fairness. Results were confirmed through simulations.

Performance Characterization for IEEE 802.11p Network With Single Channel Devices

In this paper, we investigate the performance of networks built from single-channel devices that use wireless access in vehicular environment protocols. We consider several traffic combinations, each of which presents a mix of traffic classes, over control and service channels. Our results show that time switching between the channels causes synchronization of backoff processes, which increases the frame collision probability, in particular for small sizes of contention windows. We also evaluate the impact of the interruption of the backoff process by inactive channel time, which gives rise to a probability distribution with repeated tails and a coefficient of variation larger than 1. Our model can also be used to evaluate different sets of enhanced distributed channel access parameters and to select the channel duty cycle according to the policy of the network operator.

Access delay for nodes with finite buffers in IEEE 802.15.4 beacon enabled PAN with uplink transmissions

In this work we analyze the performance of a personal area network operating under the IEEE Standard 802.15.4 in the beacon enabled mode, and derive the probability distribution of packet access delay and calculate the throughput. We assume that the network is operating in non-saturation mode and that the nodes have finite buffers. We model the operation of the PAN using the theory of discrete time Markov chains and M/G/1/K queues. The model considers acknowledged uplink transmission and includes the impact of different parameters such as packet arrival rate, number of stations, stations buffer size, packet size, and inactive period between the beacons. The model also captures the problem of congestion at the beginning of the superframe due to multiple transmissions being delayed from the previous superframe, and we propose a correction to the standard in order to avoid this problem. In order to achieve acceptable access delays and small blocking probability at the buffer (which has to be small), PAN must be operated at throughput less than 50%, which can be achieved by restricting the number of the nodes in the PAN, or by reducing the packet arrival rates at the nodes.

Bridges of Bluetooth county: topologies, scheduling, and performance

The performance of two Bluetooth piconets linked through a shared device is analyzed using the tools of queueing theory. We analyze both possible topologies: the master/slave (MS) bridge, in which the shared device is the master in one of the piconets and a slave in the other, and the slave/slave (SS) bridge, where the shared device is the slave in both piconets. Two scheduling policies, limited service and exhaustive service, are considered. Analytical results are derived for the probability distribution of access delay (i.e., the time that a packet has to wait before being serviced) and end-to-end delay for both intrapiconet and interpiconet bursty traffic. The SS bridge has been found to offer lower access delays and local end-to-end delay than its MS counterpart, which provides lower end-to-end delay for nonlocal traffic due to the smaller number of hops (three, instead of four) for such traffic. In both topologies, exhaustive service scheduling was found to provide lower delays than the limited service one. All analytical results have been confirmed through simulations.

The Effects of Test-Driven Development on External Quality and Productivity: A Meta-Analysis

This paper provides a systematic meta-analysis of 27 studies that investigate the impact of Test-Driven Development (TDD) on external code quality and productivity. The results indicate that, in general, TDD has a small positive effect on quality but little to no discernible effect on productivity. However, subgroup analysis has found both the quality improvement and the productivity drop to be much larger in industrial studies in comparison with academic studies. A larger drop of productivity was found in studies where the difference in test effort between the TDD and the control groups process was significant. A larger improvement in quality was also found in the academic studies when the difference in test effort is substantial; however, no conclusion could be derived regarding the industrial studies due to the lack of data. Finally, the influence of developer experience and task size as moderator variables was investigated, and a statistically significant positive correlation was found between task size and the magnitude of the improvement in quality.