Manghui Tu

On the optimal placement of secure data objects over Internet

Secret sharing algorithms have been used for intrusion tolerance, which ensure the confidentiality, integrity and availability of critical information. However, dynamically changing the number of shares in secret sharing schemes can be costly. To achieve performance goals in data accesses, secret sharing can be combined with dynamic replication in a distributed system with varying client access patterns. In this paper, we investigate the problem of optimal allocation of secure data objects that are secret shared and possibly replicated. The system topology we consider consists of two layers. In the upper layer, multiple clusters form a network topology that can be represented by a general graph. The nodes within each cluster also have a topology represented by a general graph. We decompose the share replica allocation problem into two sub-problems, the resident set problem, which allocates a subset of shares to clusters, and the intra-cluster allocation problem which determines the number of share replicas to be allocated and their placements. We develop two different heuristic algorithms for the two sub-problems. The algorithm for the optimal resident set problem has a time complexity of O(n/sup 2/). An O(n/sup 3/) algorithm is presented for the intra-cluster allocation problem.

Data Placement in P2P Data Grids Considering the Availability, Security, Access Performance and Load Balancing

Data dependability is an important issue in data Grids. Replication schemes have been widely used in distributed systems to ensure availability and improve access performance. Alternatively, data partitioning schemes (secret sharing, erasure coding with encryption) can be used to provide availability and, in addition, to offer confidentiality protection. In peer-to-peer data Grids, such confidentiality protection is essential since the nodes hosting the data shares may not be trustworthy or may be compromised. However, difficulties in generating new shares and potential security concerns for share reallocation make a pure data partitioning scheme not easily adaptable to dynamic user access patterns. In this paper, we consider combining replication and data partitioning to assure data availability, confidentiality, load balance, and efficient access for data Grid applications. Data are partitioned and shares are dispersed. The shares may be replicated to achieve better performance, load balance, and availability. Models for assessing confidentiality, availability, load balance, and communication cost are developed and used as the metrics to guide placement decisions. Due to the nature of contradicting goals, we model the placement decision problem as a multi-objective problem and use a genetic algorithm to determine solutions that are approximate to the Pareto optimal placement solutions.

Resilience as a New System Engineering for Cloud Computing

It has become increasingly evident that large scale systems such as clouds can be brittle and may exhibit unpredictable behavior when faced with unexpected disturbances. Even weak and innocuous disturbances can bring down the system inoperative and may introduce catastrophic disasters to the society. The goal of this research is to explore the fundamental principles and theories that govern cloud system resilience and to provide novel and effective mechanisms to model and enhance the resilience of cloud. A food web like process interaction model is developed and system resilience enhancement mechanisms are proposed based on the control of the strength of interactions. Also, the effectiveness and limitations of modularization on resilience enhancement is illustrated by using a replica consistency control protocol. The research has shown that weakening key process interactions and modularizing complex systems are very effective on resilience enhancement. 

System resilience modeling and enhancement for the cloud

It has become increasingly evident that large scale systems such as clouds can be brittle and may exhibit unpredictable behavior when faced with unexpected disturbances. Even weak and innocuous disturbances can bring down the system inoperative and may introduce catastrophic disasters to the society. The goal of this research is to explore the fundamental theories that govern cloud system resilience and to provide novel and effective mechanisms to model and enhance the resilience of cloud. A food web like process interaction model is developed and system resilience enhancement mechanisms are proposed based on the control of the strength of interactions. Also, the effectiveness and limitations of modularization on resilience enhancement is illustrated by using a replica consistency control protocol. The research has shown that weakening key process interactions and modularizing complex systems are very effective on resilience enhancement.

Maximizing the Availability of Replicated Services in Widely Distributed Systems Considering Network Availability

Availability is an important issue in distributed environments that promise to provide quality of services and thus the mechanisms to improve data or service availability is critical to ensure such quality of service. Replication schemes have been widely used to improve data or service availability. In addition to the number of replicas, the location of the replicas also impacts the data availability. In this paper, we consider the impact of node and network link failures on the availability of replicated data and services. Efficient replica allocation algorithms with runtime of O(K|V|2) are developed to improve data availability, where K is the predetermined number of replica and |V| is the number of nodes in the system. Experimental studies have been conducted to evaluate how effective the proposed replica allocation algorithms on improving the availability of replicated data or services. The results show that the two proposed solutions are effective on enhancing system availability, especially when the number of replicas is small.

Automated Generation of Integration Test Sequences from Logical Contracts

This paper exploits logical contracts (i.e., Preconditions and post conditions) to automatically generate integration tests. The approach transforms a contract-based test model into an operational model, which makes it feasible to analyze correctness of the test model and generate integration tests to meet coverage criteria of the test model. This approach has been implemented in MISTA, a tool for automated generation of executable tests. MISTA allows executable test code to be generated in a rich variety of programming languages (e.g., Java, C, C++, C#, and PHP) and test execution environments. We have evaluated the fault-detection capability of contract-based test models through fault injection. The results show that contract-based tests can be highly effective in fault detection and that robustness tests targeting the situations of broken contracts are critical to finding faults related to component interactions.

Automated Integration Testing Using Logical Contracts

Modern computer software usually consists of a number of interacting components. Defects pertaining to component interactions may not be visible until the components are integrated. To exercise component interactions, we propose an integration testing approach based on the core concept of contracts (i.e., preconditions and postconditions of component calls in first-order logic). To enable correctness analysis and generation of coverage-based tests (including state coverage, transition or component coverage, and reachability coverage) of a contract-based test model, we transform the test model into an operational model. We have implemented this approach into a framework that has the capacity to generate executable test code in a large collection of programming languages and to run the generated test code in a rich variety of test execution environments. To demonstrate the cost-effectiveness of our approach, we have conducted a series of experiments in controlled environments to measure the performance of test generation and the fault detection capabilities of coverage-based tests. The results have clearly demonstrated that: 1) our approach can quickly generate coverage-based tests when there is enough memory for dealing with the state space of the given test models; 2) reachability coverage is more capable of detecting faults than state coverage and transition coverage; 3) postcondition-based test oracles are more capable of detecting faults than state-based test oracles; and 4) robustness tests, which exercise the situations when the contracts are broken, are important for improving fault detection capability.

Reach Availability Modeling of Replicated Services

Availability is an important issue in distributed systems providing quality of services. Data or service availability modeling is very important to ensure such quality of services. Many factors, such as node availability and network link availability, could affect the data or service availability in a distributed system, and replication schemes have been widely used to improve data or service availability. In this paper, we consider node and network link failures, as well as the impact of multiple replicas on data availability. Efficient availability computing algorithms are developed to model data availability for a system with different topologies. Finally, experimental studies have been conducted to see how well our modeling approach is for the availability modeling of distributed environment with data replication.