ISH Toward a resilient software defined infrastructure to support disaster management applications

A distributed visualization system which makes it easy to understand the information generated from analysis of retrieved large-scale datasets and keeps to provide them on multiple sites concurrently is required as one of disaster management applications which support to decide and perform an appropriate decision in a disaster. For such a distributed visualization system, we aim to realize an IT infrastructure that can tolerate partial failures of components and enables to adapt to rapidly changing environments caused and/or aggravated by disaster events. The goal of this project is to prototype a Software-Defined IT infrastructure to support the continuous distributed visualization. 1. Basic Information (1) Collaborating JHPCN Centers Osaka University (2) Research Areas  Very large-scale numerical computation  Very large-scale data processing  Very large capacity network technology ・ Very large-scale information systems (3) Roles of Project Members Yasuhiro Watashiba (Nara Institute of Science and Technology (NAIST)): Representative, Resource management system José A.B. Fortes (University of Florida): Vice-Representative, Establishing international collaborations and acquiring international resources Jason Haga (National Institute of Advanced Industrial Science and Technology (AIST)): Vice-Representative, Multi-site visualization tool deployment Kohei Ichikawa (Nara Institute of Science and Technology (NAIST)): Vice-Representative, Integration of system with PRAGMA-ENT Susumu Date (Osaka University): Integration of system with PRAGMA-ENT Hirotake Abe (University of Tsukuba): Wide-area visualization testbed Yoshiyuki Kido (Osaka University): Wide-area visualization testbed Hiroaki Yamanaka (National Institute of Information and Communications Technology (NICT)): Integration of system with PRAGMA-ENT Ryousei Takano (National Institute of Advanced Industrial Science and Technology (AIST)): Wide-area visualization testbed Ryusuke Egawa (Tohoku University): Wide-area visualization testbed 2. Purpose and Significance of the Research The concept of Information-as-a-Service (Infaas) is important for constructing an efficient and flexible disaster management environment. For an appropriate decision making process in the disaster management, the flow of information to authorities and the synchronization of retrieved and analyzed data between Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures Final Report for JHPCN Joint Research of FY 2016, May 2017 2 different groups such as local governments and Non-Governmental Organizations (NGOs) must be maintained. Thus, a system which visualizes various types of data and synchronizes them between different groups is required for supporting their common understanding to make an appropriate decision in a disaster. From the consideration above, a distributed visualization system which makes it easy to understand the information generated from analysis of retrieved large-scale datasets and keeps to provide them on multiple sites concurrently is necessary for considering an appropriate decision for the disaster. Moreover, in order for the decision makers to make an appropriate decision for the disaster, the information generated from retrieved data must be continually provided for them even when the disaster makes a part of IT infrastructure on which disaster management applications are running inoperable. Since such distributed visualization system requires IT infrastructure that can tolerate partial failures of components and can adapt to rapidly changing environments caused and/or aggravated by the disaster events, we focus on the Software-Defined technology which has the potential to offer the needed flexibility and resilience to IT infrastructures. Therefore, we aim to realize a Software-Defined IT infrastructure to support the continuous distributed visualization for big data analysis during a disaster. This project is a first step in developing a large-scale crisis informatics system that is flexible and reconfigurable depending on the needs during a disaster. In this project, the functionalities of the Software-Defined IT infrastructure are assumed as the scenario shown in Figure 1. Figure 1-(A) illustrates a normal situation of distributed visualization system. If a disaster makes Site C inoperable as shown in Figure 1(B), then the resources assigned to the disaster management application are reconfigured by migrating application servers to Site B and reconstructing the assigned virtualized network slice. Figure 1 Conceptual diagram of the goal of the project. Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures Final Report for JHPCN Joint Research of FY 2016, May 2017 3 3. Significance as a JHPCN Joint Research Project To construct and evaluate the prototype of Software-Defined IT Infrastructure, we assume that it utilizes computing resources, large-scale visualization system, and network between sites capable of Software-Defined Networking (SDN). Since Osaka University has the environment to provide them, we believe that this project can be achieved by leveraging the resources. Moreover, the large-scale visualization system of Osaka University has the component technology of SDN-based distributed visualization system which is developed in the collaboration with National Institute of Information and Communications Technology (NICT). We believe that the knowledge and experience in the collaboration accelerate to achieve the Software-Defined IT infrastructure. 4. Outline of the Research Achievements up to FY 2015 (Omission. Because this project is starting