EOSC Call page: https://www.eosc-hub.eu/eosc-early-adopter-programme-2nd-call
Application form: https://eosc-hub.eu/early-adopter-programme-application-form
Submitted content: VESPA-Cloud_EOSCHub-Early-Adopter-2019_20191005_v1.pdf
Early Adopter Programme Application Form
|Provide full names with affiliations|
VESPA (Virtual European Solar and Planetary Access) is a network of interoperable data services covering all fields of Solar System Sciences. It is a mature project, developed within EUROPLANET-FP7 and EUROPLANET-2020-RI. The latter ended in Aug. 2019. It will be further supported under the EUROPLANET-2024-RI project (starting in Feb. 2020).
The VESPA data providers are using a standard API (based on the Table Access Protocol of IVOA (International Virtual Observatory Alliance) and EPNcore, a common dictionary of metadata developed by the VESPA team). The VESPA services consist in searchable metadata tables, with links (URLs) to science data products (files, web-services...). The VESPA metadata includes relevant keywords for scientific data discovery, such as data coverage (temporal, spectral, spatial…), data content (physical parameters, processing level…), data origin (observatory, instrument, publisher…) or data access (format, URL, size…). VESPA hence provides a unified data discovery service for Solar System Sciences.
The architecture of the VESPA network is distributed (the metadata tables are hosted and maintained by the VESPA providers), but it is not redundant. The hosting and maintenance of VESPA provider's servers has proved to be a single point failure for small teams with little IT support. The VESPA-Cloud project with EOSC-Hub would greatly facilitate the sustainability of data sharing from small teams as well as teams, whose institutions have restrictive firewall policies (like labs hosted by space agencies, e.g., DLR in Germany). Each VESPA data provider is using the same server software, namely DaCHS (Data Centre Helper Suite), developed by the Heidelberg team included in the project.
VESPA-Cloud will provide a cloud-hosted facility to host VESPA compliant metadata tables in a controlled and maintained software environment. The VESPA providers will focus on the science application configuration, whereas the VESPA core team will support them with the maintenance of the deployed instances. The development of the VESPA provider’s data service will be done using a git versioning system (github or institute gitlab).
An instance of the VESPA query interface portal will also be implemented on an EOSC-hub provided virtual machine.
In the course of the VESPA-Cloud project, we will implement in the DaCHS framework cloud-storage API connectors (such as Amazon S3, iRODS, etc.) to reading data in the cloud and ingesting metadata. Since DacHS is used worldwide by many datacenters to share astronomical and solar system data collections, many teams will benefit from this development.
We also propose to setup a EuroplanetResearch Community, which will include VESPA-Cloud, and other Europlanet-2024-RI and Europlanet-Societyrelated projects (such as SPIDER – Sun Planet Interaction Digital Environment on Request, or the previous services developed within PSWS – Planetary Space Weather Services).
Further development plans for VESPA-Cloud are listed below:
This new architecture is required with the growing number of services and data products served by the VESPA providers.
The VESPA providers can serve data products as well as data services (like cutout or resampling services on both data or simulation runs), or even direct calls to numerical codes through REST interfaces. The implementation of a UWS application, based on OPUS (Observatoire de Paris UWS System: https://uws-server.readthedocs.io/en/latest/) will enhance the overall interoperability between the VESPA network (and other IVOA based frameworks) with the EOSC resources.
|Please write a brief and plain explanation of the proposed project and the scientific or technical challenge it will address. Please specify how the research community will benefit from this project.|
VESPA is a mature project, with 50 VESPA providers distributing open access datasets throughout the world (EU, Japan, USA). In October 2019, the current number of data products available within the VESPA network reaches 18.3 millions (among which 5 millions products from the ESA Planetary Science Archive).
Each VESPA provider is hosting and maintaining a server (physical or virtualized) with the same software distribution (DaCHS, Data Centre Helper Suite), which implements the interoperability layers (from IVOA and VESPA) and following FAIR principles. Each server hosts a table of standardized metadata with URLs to data files or data services. Data files can be hosted by the VESPA provider team, or in an external archive (e.g., ESA/PSA - Planetary Science Archive).
The VESPA query interface portal is developed and maintained at the Observatoire de Paris (Paris, France).
A preliminary prototype of a DaCHS instance on the EOSC-hub infrastructure has been tested earlier in 2019. This instance has been ordered through the EOSC-hub marketplace (https://marketplace.eosc-portal.eu), usingEGI Cloud container compute BETA. This has been running for several weeks, with success. We could install and run the DaCHS framework, as well as serve a VESPA metadata table. This instance is now in undeployedstatus.
|Please specify what kind of technology you are using (Grids, Cloud computing, HPC, HTC, data storage, data repositories, data management systems, data discovery services, etc.) and the related Technology Readiness Level (TRL) (please, provide evidence of the declared TRL with links to technical documentation, papers, etc)|
The VESPA architecture relies on the assumption that data provider’s servers are up and running continuously. The VESPA network is distributed but not redundant. For small teams with little or no IT support is available locally, the services are down regularly. We thus need a more stable and manageable platform for hosting those services. The EOSC-hub “cloud container compute” service would solve this problem.
We propose to use the EOSC infrastructure to host VESPA provider's servers (through a controlled deployment environment with git-managed containers).
The VESPA providers would be able to:
In a second phase, we will implement an EOSC-hub hosted VESPA portal, using the web interface developed at Observatoire de Paris. The portal will be deployed from a git-based repository.
|Please explain what services and resources you are planning to use and integrate and the benefits that you are expecting from the new developed solution.|
|1.3 Physical sciences (Astronomy)||Please state the area of science of the project including one or more codes from the OECD Frascati Fields of Science most appropriate to the project, the codes can be consulted at https://www.oecd.org/science/inno/38235147.pdf|
VESPA is a distributed (although not redundant) data discovery and access framework. Hosting VESPA services in the cloud ensures their availability on the long term, and in turn the reliability of the full VESPA network. Starting in 2020, we propose to use the VESPA-Cloud infrastructure for the new VESPA providers selected after the yearly VESPA implementation workshop AO. This would add at 3 to 5 new services in 2020 from external teams. Science teams with the VESPA project will also use the VESPA-Cloud facility. We thus estimate to open about 10 services during year 2020.
The current VESPA network connects 50 data services, serving about 1.8M data products, with an average monthly visitor count of 100. With an enhanced visibility of the VESPA network through VESPA-Cloud, we expected to see a wider adoption.
VESPA-Cloud is a proof of concept, which will demonstrate the use and the efficiency of the EOSC-Hub infrastructure to the Solar System science community. Furthermore, in the course of the Europlanet-2024-RI project, the VESPA team will build strong interfaces with the planetary surfaces team building on GIS technologies (GMAP work package), as well as Space Weather (SPIDER work package) and Machine Learning (ML work package). This opens doors for reaching out similar communities focusing on Earth sciences. The VESPA-Cloud team has also existing contacts with the several ESCAPE work packages (e.g., on the interoperability, radio-astronomy, or solar physics topics).
The goal of VESPA is to make data Solar System Findable and Accessible through interoperable interfaces, and is recommending standard data and metadata formats, ensuring reusability.
All software developed for VESPA are open-source (mostly GPLv3, or Apache).
VESPA-Cloud enhances the accessibility, by providing a sustainable access for VESPA dataset
|12 months for setting up services and finding a sustainable approach for further operations.|
The minimal individual VESPA-Cloud provider’s instance is: 2 CPU, 4GB RAM, 20 GB disk.
We estimate to have about 20 instances to deploy in the first stage. Each instance must have a fixed and public IP address (customizable DNS names preferred). The instances are expected to be up and running all the time. Short unavailability of the services is acceptable, if the instance can be relaunched automatically.
Another need is a Storage Buffer for data ingestion. This is a global and temporary storage volume, which can be mounted on any VESPA-Cloud provider's instance for metadata extraction and ingestion. Data are pushed onto this volume for initial metadata extraction and ingestion and removed after this task is finished. This storage size should be 10TB. For providers needing a permanent storage capability, we will investigate with EUDAT, Zenodo or other EOSC partners.
For the current version of the VESPA portal, the compute and storage specifications are the same as for the individual VESPA-Cloud provider’s instances.
For future prospects, as listed in the project description section, it is too early to propose a sizing of needs. We will work in parallel of the VESPA-Cloud project, with the Europlanet-2024-RI work program on this infrastructure sizing.
|Please specify the minimum amount of resources needed to execute services involved in your project in terms of CPU cores, RAM, storage for data analysis|
With the new Europlanet-2024-RI program, we can expect at least 5 new providers per year, each with the same compute and storage needs.
Among the further developments foreseen after the completion of the pilot program:
However, they are not yet not completely defined and sized, since this is part of the work to be accomplished in Europlanet-2024-RI.
|Please specify the amount of resources in terms of CPU cores, RAM, storage for data analysis needed to scale-up your Project after the completion of the pilot|
Not applicable, see below.
|Please describe your data sets in volume and granularity and specify the minimum amount of archive storage as well as the access requirements for long-term archiving|
|The VESPA-Cloud services are hosting metadata tables of data products hosted on independent facilities. Each metadata table can be rebuilt from a resource descriptor script maintained with a git versioning system. The archiving of the computed metadata tables is not planned yet, but any discussion on this specific point is welcome. The archiving of the resource descriptor script should also be discussed, but it may contain private data (such as logins and tokens), which shall not be openly available. The provider's teams are currently in charge of their own data preservation, outside of the VESPA or VESPA-Cloud projects. However, the VESPA-Cloud provider’s instances are expected to run as long as the data providers are sharing their data. A long-term sustainability plan has to be prepared during the Pilot program, together with the new-born Europlanet Society (https://www.europlanet-society.org).||Please specify the amount of storage for long-term archiving and corresponding data policies|
|No sensitive data.||Is the data to be handled Regulated data, Privacy-sensitive etc.?|