Architectural recovering model for Distributed Databases: A reliability, availability and serviceability approach

• Published in: 2016 IEEE Symposium on Computers and Communication (ISCC) pp. 575 - 580
• Main Authors: Hugo de Souza, Ramon, Flores, Paulo Arion, Ribeiro Dantas, Mario Antonio, Siqueira, Frank
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-06-2016
• Subjects: Availability; big data; cloud services; Computers; distributed database architectures; Mathematical model; Measurement; quality of experience; Quality of service; Reliability; serviceability; Time factors
• DOI: 10.1109/ISCC.2016.7543799

The early theoretical studies that presented conceptual definitions for the mapping of Quality of Experience (QoE) through Quality of Service (QoS), in an approach focused on Distributed Databases (DDBs), showed the directions to be threshed in a way to accomplish a hybrid DDBs' QoE-QoS set of evaluation criteria. That evaluation set being classified explicitly as partial, since QoE is a known multidimensional concept, hard to be explicitly well-defined. With these newfound equations being useful specifically to system-level purposes, without focusing on the user explicitly, as the usual QoE approach, these studies presented a new abstraction that allows web storage cloud services to adequately explore and assess requirements for contracted services. This study presents a novel architectural model to deal with reliability, availability and serviceability issues, in order to provide a solution giving QoS-like guarantees to DDB systems. With its origins in this set of evaluation criteria, this new kind of architectural solution aims to provide more stability by dealing with statistical guarantees. The proposed architecture focuses on these evaluations to choose recovery procedures, in order to avoid future unexpected behaviour, dealing with horizontal and vertical DDB's solutions to keep the services according to a Service Level Agreement (SLA).