INDIGO-DATA CLOUD EC project: A study case applied to one of the EMSO Research Infrastructure Deep sea Observatories

摘要

Our case study is a pilot experience used to describe some of the activities performed by INGV in the frame of the European Research Infrastructure EMSO (European Multidisciplinary Seafloor and water column Observatory). EMSOis composed of several deep-seafloor and water column observatories, deployed at key sites in the European waters, thus forming a widely distributed pan-European infrastructure. We consider data collected by the NEMO-SN1observatory, one of the EMSO nodes used for geohazard monitoring, located in the Western Ionian Sea in proximity of Etna volcano. In this poster we will focus on the Researcher and Data Manager user-types. The INGV EMSOcommunity uses MOIST (Multidisciplinary Oceanic Information System) for storing and visualizing data and metadata produced by NEMO-SN1 Observatory. Data quality control and analysis often requires several steps that include theuse of different scripts and software developed in-house, commercials tools (Matlab, R-Studio....), and proprietary tools available from sensor manufacturers. In this chain of events, some operations might require a relevant computingpower. Data are retrieved from MOIST through remote mount (via samba or sshfs). Analysis might also be performed on datasets that are produced by other partners and remote access and sharing of these data is needed. At present,in the majority of cases, software is run on individual researchers’ Pcs. The first test for the implementation of our use-case within INDIGO-DataCloud included running an R script on a cloud environment and exploiting data sharingcapabilities. The input to this script is data coming from the analysis of Short Duration seismic Events (SDE) automatically detected on the seismometer continuous time series. The script calculates a cumulate energy in themeasurement period (8 months) and compares this cumulate curve to N random cumulates calculated by mixing the energy values at the fixed observed times where SDE are detected. Within the INDIGO project (WP2) we defined theusers requirements and we identified some useful INDIGO solutions. In particular we are testing Ophidia, a software stack for big data analytics (Fiore et al., 2013), and the execution of R jobs in docker containers described by TOSCAtemplates through INDIGO Orchestrator and Apache Mesos. This poster illustrates our case study, the users’ requirements and the INDIGO solutions that we have been testing so far and would like to test in the near future.