Protecting databases from schema disclosure: a CRUD-based protection model

Database schemas, in many organizations, are considered one of the critical assets to be protected. From database schemas, it is not only possible to infer the information being collected but also the way organizations manage their businesses and/or activities. One of the ways to disclose database schemas is through the Create, Read, Update and Delete (CRUD) expressions. In fact, their use can follow strict security rules or be unregulated by malicious users. In the first case, users are required to master database schemas. This can be critical when applications that access the database directly, which we call database interface applications (DIA), are developed by third party organizations via outsourcing. In the second case, users can disclose partially or totally database schemas following malicious algorithms based on CRUD expressions. To overcome this vulnerability, we propose a new technique where CRUD expressions cannot be directly manipulated by DIAs any more. Whenever a DIA starts-up, the associated database server generates a random codified token for each CRUD expression and sends it to the DIA that the database servers can use to execute the correspondent CRUD expression. In order to validate our proposal, we present a conceptual architectural model and a proof of concept.

Arquitectura e plataforma de Middleware para suporte de bibliotecas digitais distribuídas e ecléticas

O trabalho apresentado nesta dissertação teve por objectivo principal a concepção, modelação e desenvolvimento de uma plataforma de middleware que permitisse a integração de sistemas de informação, em todos os seus níveis (dados, lógico e apresentação), perfazendo uma federação de bibliotecas digitais distribuídas e ecléticas. Para este fim, foram estudadas as várias abordagens de modelação e organização das bibliotecas digitais, assim como os diversos sistemas e tecnologias de suporte existentes no momento inicial do trabalho. Compreendendo a existência de muitas lacunas ainda neste domínio, nomeadamente ao nível da interoperabilidade de sistemas heterogéneos e integração da semântica de metadados, decidiu-se proceder a um trabalho de investigação e desenvolvimento que pudesse apresentar eventuais soluções para o preenchimento de tais lacunas. Desta forma, surgem neste trabalho duas tecnologias, o XML e o Dublin Core, que servem de base a todas as restantes tecnologias usadas para a interoperabilidade e para a integração. Ainda utilizando estas tecnologias base, foram estudados e desenvolvidos meios simples, mas eficientes, de salvaguarda, indexação e pesquisa de informação, tentando manter a independência face aos grandes produtores de bases de dados, que só por si não resolvem alguns dos problemas mais críticos da investigação no domínio das bibliotecas digitais. ABSTRACT: The main objective of the work presented in this dissertation is the design, modulation and development of a middleware framework to allow information systems interoperability, in all their scope (data, logic and presentation), to accomplish a distributed and eclectic digital libraries federation. Several modulations and organizations were approached, and several support systems and technologies were studied. Understanding the existence of many gaps in this domain, namely in heterogeneous information systems interoperation and metadata semantic integration, it was decided to conduct a research and development work, which, eventually, could present some solutions to fill in these gaps. In this way, two technologies, XML and Dublin Core, appear to serve as the basis of all remaining technologies, to interoperate and to achieve semantic integration. Using yet these technologies, it was also studied and developed simple means, but efficient ones, to save, index and query information, preserving the independence from major data base producers, which by their selves don’t solve critical problems in the digital libraries research domain.

DACA: architecture to implement dynamic access control mechanisms on business tier components

Access control is a software engineering challenge in database applications. Currently, there is no satisfactory solution to dynamically implement evolving fine-grained access control mechanisms (FGACM) on business tiers of relational database applications. To tackle this access control gap, we propose an architecture, herein referred to as Dynamic Access Control Architecture (DACA). DACA allows FGACM to be dynamically built and updated at runtime in accordance with the established fine-grained access control policies (FGACP). DACA explores and makes use of Call Level Interfaces (CLI) features to implement FGACM on business tiers. Among the features, we emphasize their performance and their multiple access modes to data residing on relational databases. The different access modes of CLI are wrapped by typed objects driven by FGACM, which are built and updated at runtime. Programmers prescind of traditional access modes of CLI and start using the ones dynamically implemented and updated. DACA comprises three main components: Policy Server (repository of metadata for FGACM), Dynamic Access Control Component (DACC) (business tier component responsible for implementing FGACM) and Policy Manager (broker between DACC and Policy Server). Unlike current approaches, DACA is not dependent on any particular access control model or on any access control policy, this way promoting its applicability to a wide range of different situations. In order to validate DACA, a solution based on Java, Java Database Connectivity (JDBC) and SQL Server was devised and implemented. Two evaluations were carried out. The first one evaluates DACA capability to implement and update FGACM dynamically, at runtime, and, the second one assesses DACA performance against a standard use of JDBC without any FGACM. The collected results show that DACA is an effective approach for implementing evolving FGACM on business tiers based on Call Level Interfaces, in this case JDBC.