EUXDAT offers both development and operational environments for agriculture-oriented service providers on top of a large set of data from satellites, UAV, in-situ sensors, meteorological data, land-use and soil. Concerning satellite data, Sentinel products are hosted and provided by Mundi DIAS on top of which EUXDAT e-Infrastructure is built.
When you develop a processing service on EUXDAT, you can access Sentinel data through several modes, depending on the way this data is used in your service : Do you need the raw bands of the sentinel products or preprocessed subsets of the products? Do you need to start from the full Sentinel-tile or just a crop on a specific area ? Is the needed format an image, a layer to display on top of previous images, raw data or statistical information ? Do you need to compute custom indices from a combination of the raw bands ?
The size of a single Level-1C Sentinel2 product with a 100kmx100km resolution is around 600MB, and it reaches 800MB for a Level-2A, so the question is often: should I download the product and preprocess it on my own in order not to store large products? Or should I download the preprocessed sub-part of the product directly, with lower storage needs but sometimes lower processing performances? There is no unique answer to this question, but let’s have a look at the different options.
In order to search for all the products corresponding to both a time and area of interest and to download the raw bands of the product in GeoTIFF or jpeg2000 format, a REST API is available.
From Jupyter notebooks and more generally Python code, a
python library is available to provide the search, download , scale and display
functionalities, along with additional features : search using OpenSearch
standard, access OGC services : CSW (Catalogue Service for the Web), WCS (Web
Coverage Service, WFS (Web Feature
Service), WMS (Web Map Service), WMTS
(Web Map Tile Service).
This library can be useful for the search and download functionalities but has two important limitations for OGC WMS and WCS services : the resolution parameter cannot be used so the resolution is fixed at 100kmx100km, and the area of interest cannot be a geometry, only a rectangle (bbox) which is an important restriction when the element of interest is a parcel.
When the need is to access preprocessed data rather than raw product, it is possible to let a preprocessing engine handle these preprocessing steps, and to download directly the result of the preprocessing rather than the Sentinel product and all the intermediate steps.
Sentinel Hub is such an engine and is deployed on Mundi. All OGC endpoints (WMS, WFS, WCS, WMTS) available from Mundi comply with both OGC standard operations and additional Sentinel Hub custom parameters  for advanced filtering (max cloud coverage) or definition of the area of interest as a polygon rather than a rectangle. These OGC endpoints can be used in EUXDAT either directly (HTTP request) or through Sentinel Hub Python API . A very interesting feature concerns the ability to use custom script to combine raw bands, for example in the case of vegetation indices, and to obtain directly the preprocessed WMS or WCS result of band combination, cropped on an area of interest. The example below shows a WCS request using Sentinel hub Python API from a Jupyter notebook on EUXDAT and defines a custom evaluation script for the Leaf Area index calculation (LAI Gitelson), computed from raw bands as B08/B05-1, on a polygon area, resulting in a GeoTIFF file.
To go further, the EOLearn  python library can be used to define and run preprocessing operations as a workflow. Each preprocessing step is defined as an EOTask and run on an EOPatch that corresponds to the result of a request on the Sentinel data source. This interesting approach enables both the externalization of the preprocessing steps to an external engine and the production of time-series results.
EUXDAT offers these different approaches to access Sentinel data: from REST API to Python library, from raw data access and download to the result of a preprocessing workflow with Sentinel Hub or EOLearn libraries. It is up to you to choose the mode that suits your needs.
Sentinel Hub and EOLearn libraries are also an important part of the upcoming “European Data cube Facility service” project  for the European Space Agency, which aims at providing an easier access to EO data, in an “analysis-ready” format, as an operational service for EO platform. With this future improved access to the EO data, the capabilities of processing services for EUXDAT or similar EO platforms will probably increase in the future months or years.
 SentinelHub custom parameters : https://www.sentinel-hub.com/develop/documentation/api/custom-url-parameters
 SentinelHub python library reference : https://sentinelhub-py.readthedocs.io/en/latest/
 European data cube facility service project https://eo4society.esa.int/2019/05/21/european-data-cube-facility-service-an-eo-resource-factory/