Toshiyuki Miyachi

StarBED and SpringOS: large-scale general purpose network testbed and supporting software

New technologies for the Internet should be evaluated on environments dedicated to experiments, in order to avoid influences to critical services on the Internet. Generally software simulation and small testbed using real nodes are used to satisfy these requirements. There are several stages in developing new technologies, however, and these technologies may not satisfy requirements for all stages.We pointed the gap between the Internet and environment for experiment, especially in aspects of scale, complexity and reality. We proposed and implemented StarBED which is a testbed based on lots of actual nodes, in order to build large-scale, complex and realistic environments for experiments. StarBED consists of 512 PCs and switches which connect these PCs. The PCs on StarBED are designed to run 10 virtual PCs on a physical PC. It enables to build a topology for experiments using up to 5120 nodes. It is difficult to manage and control such a lot of nodes. We also designed and implemented SpringOS, which is a supporting software for making experiments. SpringOS makes the topology and drives the scenario for experiment according to the users configuration automatically.Many experiments were performed on StarBED, and this shows StarBEDs effectiveness.

Automatic configuration and execution of Internet experiments on an actual node-based testbed

Software simulators are widely used for validation and evaluation of new network technologies and services. Using software simulators is a good way to validate algorithms or observing microbehavior of communication protocols. There are, however, problems with software simulators. Most software simulators require target systems to be described under their own modelling scheme, often using their own modelling language. These descriptions are usually different from what will actually be running as products. It is clear that these products should be validated someway. Time required to run software simulation become problematic also, as we try to simulate realistic target system under realistic environment where nontrivial aggregation of complex network services come into play. We adopt an approach to prepare a configurable testbed using actual nodes. Experiment topologies are created on this testbed virtually without changing physical connections, because the cost of building such experiment environments is very large. Since users of such testbed have to perform many steps to execute the desired experiments on such environment, we design the system that supports the users to execute their experiments. Using our system, all the user have to perform is preparing a experiment configuration file. Our system will execute experiments according to the configuration file. This paper shows the design of our supporting system models, steps of experiment with our system and an example of users scenario.

QOMB: A Wireless Network Emulation Testbed

In this paper we present QOMB, a testbed we designed and implemented for the evaluation of wireless network systems, protocols and applications. The testbed uses the wireless network emulation set of tools QOMET so as to reproduce in a wired network, in real time, the wireless network conditions corresponding to a given scenario. In this context QOMET also provides support for features such as realistic virtual 3D environments, and node mobility generation. The infrastructure of QOMB is StarBED, the large-scale network experiment testbed at the National Institute of Information and Communications Technology, Hokuriku Research Center, in Ishikawa, Japan. The multi-hop wireless network emulation experimental results related to OLSR performance analysis in mesh networks and MANETs illustrate the main features and the usability of QOMB.

Experiences in emulating 10K AS topology with massive VM multiplexing

New technologies that will be introduced to the Internet should be practically tested for effectiveness and for side effects. A realistic environment that simulates the Internet is needed to experimentally test such technologies, which will be widely deployed on the Internet. To support experimentation in a realistic, Internet-like environment, we are now trying to construct an Internet on a testbed. We describe our method of constructing an Internet-like environment on the testbed using a virtualization technology and estimation of the inter-AS network on StarBED with Xen and our prototype system. We stably constructed a 10,000-AS network using 150 testbed nodes and estimated its performance and feasibility.

Design issues of an isolated sandbox used to analyze malwares

Recent viruses, worms, and bots, called malwares, often have anti-analysis functions such as mechanisms that confirm connectivity to certain Internet hosts and detect virtualized environments. We discuss how malwares can be kept alive in an analyzing environment by disabling their anti-analyzing mechanisms. To avoid any impacts to/from the Internet, we conclude that analyzing environments should be disconnected from the Internet but must be able to make malwares believe that they are connected to the real Internet. We also conclude that, for executing environments to analyze anti-virtualization malwares, they should not be virtualized but must be as easily reconstructable as a virtualized environment. To reconcile these cross-purposes, we propose an approach that consists of a mimetic Internet and a malware incubator with swappable actual nodes. We implemented a prototype system and conducted an experiment to test the adequacy of our approach.

Educational environment on StarBED: case study of SOI Asia 2008 Spring Global E-Workshop

Hands-on experiences are indispensable to train IT operators. However, it is often difficult to conduct hands-on practices in many locations due to lack of resources such as PCs and network equipments. The cost of gathering lecturers and participants at one hands-on site is another possible problem in this situation. Utilization of remotely located hands-on environment is one of the solution to solve these problems. StarBED is a testbed facility for conducting network experiments. Installation of any OSes, adjusting the network topologies according to the requirements can be easily realized utilizing StarBED and SpringOS. A software suite, SpringOS, was developed exclusively for automating the experiment setups in StarBED. This facility could be used to address the problems of lack of resources and high cost for traveling. This research conducted a region-wide remote hands-on workshop utilizing the StarBED inviting 42 participants from 10 Asian countries. The workshop was designed so that the remote participants do not have to travel to single place, and prepare same specification PCs at each site. Through the collected data and feedback, the proposed workshop model was proved to be feasible and effective. This paper describes the requirements and approaches for region-wide remote hands-on workshop utilizing StarBED for building target hands-on environment. As the contents of the conducted workshop could be applied to the general IT human resource development programs, this paper could be useful for the future remote hands-on programs to train IT operators.

A rendezvous in network experiment-case study of Kuroyuri

Rendezvous among experiment entities appear in network experiments both implicitly and explicitly. When the driving system for a network experiment testbed supports such rendezvous, users are free to focus on the content of the experiments, without having to consider details of the rendezvous (conditions, timing, etc.) and their execution. This paper shows typical entity activities in network experiments, especially rendezvous, and presents our approach in the Kuroyuri network experiment driving program

StarBED2: Testbed for Networked Sensing Systems

Nowadays many new technologies are being developed and introduced for Internet, home networks, and sensor networks. The new technologies must be evaluated in detail before deployment. However the above mentioned networks have a large number of nodes, and a complicated topology. We developed a large-scale, realistic and real-time network testbed, StarBED, using hundreds of PCs, and switched networks. We are now implementing StarBED2, which expands StarBED so as to be suitable for emulating ubiquitous networks. In this paper we describe StarBED2, its design policy, architecture, and additional components, including the custom experiment- support system, RUNE (real-time ubiquitous network emulation environment). We then show some experimental results obtained in this environment with emulated networked sensing systems.

XBurner: A XENebula-Based Native Traffic-Generation Platform

There are two types of network traffic in experimental environments. One is traffic which is derived from target elements and the other one is background traffic which is derived from surrounding elements. There is very little knowledge on the relationships between new elements and existing elements; therefore, surrounding elements that seem to have no apparent relationship with the target elements should also introduced into the environmental environments. Therefore emulating background traffic is important to take realistic experimental results.

Emulation-Based ICT System Resiliency Verification for Disaster Situations

Disaster situations require resilient ICT systems in order to provide as good as possible communication conditions in such catastrophic circumstances. The resiliency verification of ICT systems is however difficult, because reproducing large-scale disaster conditions in production networks is impossible without affecting their users. In this paper we present an emulation-based approach for evaluating the resiliency of ICT systems in disaster situations. This is achieved by reproducing disaster-like conditions in a emulated environment running on a large-scale test bed on which actual network protocols and applications are executed in real time. The experimental results presented in the paper demonstrate how the effects of both the emulated disasters and those of the recovery technologies we subsequently deploy can be quantified objectively. This methodology can be used to improve the resilience of the systems under test.