Javam C. Machado

Towards Elastic Multi-Tenant Database Replication with Quality of Service

Database systems serving cloud platforms must handle a large number of applications or tenants. Multi-tenancy database has been prevalent for hosting multiple tenants within a single DBMS while enabling effective resource sharing. Providing such performance goals is challenging for providers as they must balance the performance that they can deliver to tenants and the operating costs. In such database systems, a key functionality for service providers is database replication, which is useful for availability, performance, elasticity, and quality of service. This paper presents RepliC, an approach to database replication in the cloud with quality of service, elasticity, and support tomulti-tenancy. In order to evaluate RepliC, some experiments that measure the quality of service and elasticity are presented. Our experiment results confirm that RepliC ensures the quality of service with small SLA violations, while using resources efficiently.

Scale-Space Filtering for Workload Analysis and Forecast

Dynamic resource provisioning poses a major challenge for infrastructure providers because it is necessary to both forecast resource consumption and react to recent surges on demand for maintaining a tradeoff between quality of service and cost. However, approaches to workload analysis and forecast are affected due to noise in observed data, specially in forecast models. Moreover, most studies do not consider different prediction horizons may be necessary in order to take action before an SLA violation occurs. This paper presents an approach based in the scale-space theory to assist the dynamic resource provisioning. This theory is capable of eliminating the presence of irrelevant information from a signal that can potentially induce wrong or late decision making. In order to evaluate our approach, some experiments are presented considering both reactive and proactive strategies. The results confirm that our approach improves the workload analysis and forecast.

A distributed concurrency control mechanism for XML data

XML has become a widely used standard for data exchange among applications. Consequently, a large amount of data is distributed on the Web and stored in different persistence models. DBMSs provide concurrency control techniques to manage such data. However, the structure of XML data makes the application of these techniques difficult. Regarding distributed environments, there are few papers available and they all have limitations. This paper introduces DTX, a mechanism for distributed concurrency control for XML data. In order to evaluate DTX, experiments that measure its performance are presented.

A*-based Solutions for KNN Queries with Operating Time Constraints in Time-Dependent Road Networks

We consider the problem of finding the k nearest points of interest from a given location in time-dependent road networks, i.e., One where travel time along each edge is a function of the departure time, and where the operating times of the points of interest are also taken into consideration. More specifically, we address the following query: find the k points of interest in which a user can start to be served in the minimum amount of time, accounting for both the travel time to the point of interest and the waiting time, if it is closed. Previous works have proposed solutions to answer kNN queries considering the time dependency of the network but not the operating times of the points of interest. We propose and discuss three solutions to this type of query which are based on the previously proposed incremental network expansion and use the A search algorithm equipped with suitable heuristic functions. We also present experimental results comparing the number of disk access required in each solution with respect to a few different parameters.

Nearest neighbor queries with service time constraints in time-dependent road networks

We consider the problem of finding the closest point of interest in road networks where the travel time along each edge is a function of the departure time and the service time of the facilities are taken into consideration. More specifically, our query aims to find the point of interest (e.g., a museum) in which the user can be served in the minimum amount of time, accounting for both the travel time to the point of interest and the waiting time, if it is closed. We discuss our proposed solution and present experimental results comparing the quality of the response returned by our proposed algorithm to an approach where possible waiting times are not considered.

The light pollution as a surrogate for urban population of the US cities

Abstract We show that the definition of the city boundaries can have a dramatic influence on the scaling behavior of the night-time light (NTL) as a function of population (POP) in the US. Precisely, our results show that the arbitrary geopolitical definition based on the Metropolitan/Consolidated Metropolitan Statistical Areas (MSA/CMSA) leads to a sublinear power-law growth of NTL with POP. On the other hand, when cities are defined according to a more natural agglomeration criteria, namely, the City Clustering Algorithm (CCA), an isometric relation emerges between NTL and population. This discrepancy is compatible with results from previous works showing that the scaling behaviors of various urban indicators with population can be substantially different for distinct definitions of city boundaries. Moreover, considering the CCA definition as more adequate than the MSA/CMSA one because the former does not violate the expected extensivity between land population and area of their generated clusters, we conclude that, without loss of generality, the CCA measures of light pollution and population could be interchangeably utilized in future studies.

A new approach to replication of XML data

XML has become a widely used standard for data exchange in several application domains. In order to manage data in this format, Native XML Databases (NXDBs) are being proposed and implemented. Even though, currently there are a number of available NXDBs, few of them provide replication mechanisms. This paper presents RepliX, a mechanism for replication of XML data based on group communication. With the purpose of validating RepliX, experiments were conducted to measure its performance.

DMM: A distributed map-matching algorithm using the MapReduce paradigm

Map-matching is the problem consisting of matching a sequence of geographic coordinates with the roads on a digital map, aiming to discover the actual path traveled by that trajectory. Common uses of map-matching include traffic analysis and flow density extraction, which rely on such algorithms as a primary stage of their processing. Here we present DMM, a distributed solution for large-scale trajectory data processing and suitable for low-sampling-rate GPS trajectories. This solution is based on another low-sampling algorithm [1], and adapted to work in a distributed manner, using the MapReduce paradigm. For this purpose, we heavily rely on the Apache Spark framework and its data abstraction, Resilient Distributed Datasets (RDDs). The experiments show that the DMM algorithm has high accuracy and scalability and can be used for trajectory data streams.

Machine Learning Approach for Cloud NoSQL Databases Performance Modeling

Cloud computing is a successful, emerging paradigm that supports on-demand services with pay-as-you-go model. With the exponential growth of data, NoSQL databases have been used to manage data in the cloud. In these newly emerging settings, mechanisms to guarantee Quality of Service heavily relies on performance predictability, i.e., the ability to estimate the impact of concurrent query execution on the performance of individual queries in a continuously evolving workload. This paper presents a performance modeling approach for NoSQL databases in terms of performance metrics which is capable of capturing the non-linear effects caused by concurrency and distribution aspects. Experimental results confirm that our performance modeling can accurately predict mean response time measurements under a wide range of workload configurations.

Predicting Overtemperature Events in Graphics Cards Using Regression Models

Graphics cards are complex electronic systems designed for high performance applications. Due to its processing power, graphics cards may operate at high temperatures, leading its components to a significant degradation level. This fact is even more present when any of the heat exchange components is not working properly. In such cases, graphics cards may operate in temperatures that are higher than the specified by the manufacturers. This work presents a methodology to detect over temperature events in graphics cards using regression models. The proposed approach was tested in real graphics cards from different manufacturers. The final model achieved promising results.