Idris Winarno

Simulating Resilient Server Using XEN Virtualization

Abstract Since servers play a critical role in data processing and data transmission for serving many clients, failures in servers cause not only performance degradation of the server itself but also threats to all the computers connected to the server. Resilient servers that can self-recognize failures, self-repair failures and self-replace failed parts are required when computer systems and networks become such huge-scale as witnessed by data centers and the cloud computing. As a preliminary study, we report simulations restricting ourselves to demonstrate the self-repair network (SRN) model in a homogeneous environment realized by a virtualization technique. Simulations are conducted using native and hosted VMM on a single physical computer. Three scenarios: hang faulty, DoS attack and virus infection, are simulated. These simulations demonstrated how a server with a homogeneous environment (realized by the self-repair network model using the virtualization technique) can or cannot keep the resilience, and even suggested a possibility and necessity of using a self-reconfiguration model to create diversity.

A Cyber Attack-Resilient Server Using Hybrid Virtualization

This paper describes a novel, cyber attack-resilient server using hybrid virtualization that can reduce the downtime of the server and enhance the diversity of operating systems by adding a Linux virtual machine. The hybrid virtualization consists of machine- and application-level virtualization. The prototype system virtualizes a machine using VMware ESXi, while the prototype system virtualizes a server application using Docker on a Linux virtual machine. Docker increases the speed at which a server application starts while requiring fewer resources such as memory and storage. Performance tests showed that the prototype system reduced the downtime of the DNS service by exploiting a vulnerability with no false positive detections compared with our previous work.

Increasing the Diversity of Resilient Server Using Multiple Virtualization Engines

The Virtual machine monitor and container is an example of virtualization engine technologies that are currently in wide use. Both types of virtualization engines have vulnerabilities that could hamper a virtual machine. In our previous work, we created a resilient server using virtualization technologies with a single virtualization engine. To improve the resilience of the server, we use two different types of virtualization engines that monitor and respond by the SRN manager. The experiments results show that using multiple virtualization engines could cover their vulnerabilities against limited scenarios of failure.

Towards an immunity-based approach for preserving energy of data-gathering processes in wireless sensor network nvironments

Wireless sensor networks (WSN) consists of large number of sensor nodes with limited energy and computing resources. These facts have highlighted open problem to prolong the lifetime of sensor nodes while still maintaining the functional roles of WSN in reporting important events to a base station. These papers propose an approach for WSN inspired by the concept of immunity-based system where a set of reactions is performed in the appearance of nonself substances called syndromes. The notion of reactive cluster is introduced where only sensor nodes affected by the syndrome appearances need to perform reactions by establishing clusters among them. Since communication costs significant energy, each sensor nodes as an immune cell need to reduce impacts of the syndromes for itself and other sensor nodes in the routing paths to the sink node. It is done establishing adaptive data aggregation strategies. Depend on the critical level of the corresponding syndrome and error of wireless communications that may occur, the length of the aggregation window is adjusted. We show that the proposed approach shows promising results in minimizing dead sensor nodes, avoiding network congestion while still maintaining the functional roles of WSN in reporting important events to the sink node.

Simulating resilient server using software-defined networking

Computer (server) becomes an essential part of the information system. However, this server always has a possibility to experience a failure. In our previous study, we involve virtual machine technologies to address the failure that occurs on the server. However, this method only covers the recovery of the server from the perspective of the server itself. Our proposed method is to use the concept of software-defined networking to overcome the failure by involving network devices (e.g., switches). The experiment result shows that the concept of SDN is able to maintain the resilience of the server by interrupting the suspicious traffic.