The following is copied from OAIS v1 (2002) and may be out of date

1 ARCHIVE SCENARIO FOR THE CENTRE DES DONNEES DE LA PHYSIQUE DES PLASMAS (CDPP)

1.1 DOMAIN AND CUSTOMERS

The CDPP (Centre des Données de la Physique des Plasmas - Center for Data on Plasma Physics) is a new service currently being set up. It has been developed to ensure the long-term conservation and availability of natural Plasma Physics data (magnetospheric plasma, planetary plasma, etc.) for the international scientific community. More specifically, the data concerned is from either ground-based or space-flown experiments in which France has participated or wholly directed. The CDPP is designed around two principal components:

– Technical Activity segment, located on the premises of the French space agency, CNES, mainly in charge of developing and maintaining the archive system. The latter has the following functions: addition of data and metadata to the system, preservation of data and metadata, organization of search and product ordering facilities, and dissemination.

– Scientific Activity segment, located at the CESR (Centre d'Etudes Spatiales des Rayonnements - Center for the Study of Space Radiation), a science laboratory near CNES. The CESR is in charge of all aspects relating to scientific knowledge of the data: validating data with its producers, ensuring that the data is useable by the scientific community, setting up added-value services, etc. This Center is also responsible for developing a WWW server to present CDPP services, supplying educational information on Plasma Physics to the general public, and guiding users to access and dissemination functions.

The two complementary segments work closely together.

A number of associated laboratories will be able to join the two main components of the CDPP provided they offer a service (data dissemination or information) relating to natural plasma physics.

The archive system is currently being validated. The service is planned to be made available to the scientific community on September 1999.

Data Producers. Data producers are mainly either current or future experiments, or projects concerned with rehabilitating existing data. Ongoing experiments include, for example, the French experiments flown aboard Russian satellites (INTERBALL), aboard the US satellite (WIND), aboard the future European satellites (CLUSTER2), or even some data from the EISCAT radars. The projects to rehabilitate existing data cover a many French experiments performed since 1975, mostly flown on European, US and Soviet satellites or probes.

1.2 INGEST

The CDPP has drawn up a specification for deliverable data products. The specification defines the characteristics (either mandatory or optional) that the data and metadata to be delivered to the CDPP must exhibit. It defines the rules systematically applied with respect to:

– file structure, data encoding and standardization of times and dates;

– orbit or trajectory data;

– the minimum content and format of catalogues;

– complementary information needed to use or interpret the data.

The CDPP provides technical support in order to apply this specification to each data-producing project.

As far as future projects are concerned, the authorities empowered to make decisions on projects will make the drawing up of an obligatory data management plan. The plan must define exactly which data will be archived (physical values, raw data, etc.), how the data will be organized, and when the data will be delivered to the CDPP.

One particular service within the CDPP is the G2ID (Groupe de Gestion des Informations et des Données - Service for Managing Information and Data), in charge of the interfaces with data-producing projects and the formatting of some metadata before its delivery to the archive system.

Submission Agreements. As far as future projects are concerned, the submission agreement shall be constituted by the project Data Management Plan, to be approved by both the project and the CDPP. As far as existing data to be rehabilitated is concerned, the framework is less formal; there is normally no project team left and no longer a budget specific to that project. Rehabilitation is thus the responsibility of a team of engineers from CNES and those of the Principal Investigator or members of his team. The CDPP suggests priorities for the work to be completed. It also influences the choices and compromises to be made with regard to the level of data to be archived.

Delivery Session

Data delivery. Data-producing projects must normally store the data produced before delivery. They do so using the facilities offered by the STAF, a multi-mission storage service at CNES. The main function of the STAF (Service de Transfert et d'Archivage des Fichiers - Service for Transferring and Archiving Files) is the long-term physical storage of files. The interface is stable and therefore the technologies and storage media can thus be replaced or changed in-house without affecting the interface. The STAF also monitors and renews the media used.

From a user project viewpoint, the STAF appears as a virtual tree structure in which files may be stored. When all the data to be delivered has been produced, the delivery process merely amounts to a change of ownership of the STAF directories in which the data is stored. There is no actual physical movement of data.

Delivery of metadata. Metadata generally takes up less space than data. A delivery disk space is set up by the CDPP and the data-producing project has the right to write onto this space. When all the data and metadata has been delivered, the G2ID can begin its checking and formatting (see below). This process is valid for a complete set of data, a partial delivery or an update of previously delivered metadata.

Transformation Process

The format of experiment data is not altered during the delivery process. On the other hand, metadata delivered will be subject to a kind of packing (without changing the contents), and new metadata will be created by the G2ID. To give some examples:

– The archive system manages the descriptions of both data collections and objects, browse data and documentary information in the form of graphs on collections and objects. The delivery of a new collection results in the creation of a new node in the data description graph and logical links with existing collections. The creation of this information, granting a global and consistent view of all the data and metadata available, is not within the domain of the data producer.

– When the Principal Investigator delivers a Microsoft Word document describing an experiment, he places the corresponding file in the delivery disk space. The G2ID will then use this file to create a documentary object descriptor giving the document title, author, publishing body, language, associated keywords, stating the existence of an abstract, etc.

The insertion of metadata in the archive system is mostly based on use of Parameter Value Language (PVL) and a Data Entity Dictionary (DED), which is configuration managed. One of the roles of the G2ID will thus be to create this new metadata and construct the PVL structure describing it. Generally speaking, metadata appears as an extremely heterogeneous set of information objects. Using PVL means that these heterogeneous objects may be delivered in both a homogeneous and standard format.

Validation

The G2ID is responsible for ensuring that the deliverable product specifications for each data set have been respected. It also performs a number of coherence checks, such as checking coherence between catalogue data and the files containing experiment data.

Once these checks have been completed, the results, together with all the metadata, are presented at a formal peer review whose purpose is to decide whether the CDPP can accept the data set and issue recommendations in this field. Once accepted, the CDPP becomes the guarantor of the data set. This review brings in scientists from outside both the CDPP and the Principal Investigator team.

Despite the various checks carried out, the scientific validity of the experiment data delivered to the CDPP remains the responsibility of the Principal Investigator or data-producing project.

Security. The delivery process for both data and metadata takes place within a dedicated environment, accessible only by the data producer and the CDPP.

1.3 INTERNAL FORMS

Storage. The STAF multi-mission storage service (see above) takes charge of the data and metadata. This service currently uses several StorageTek silos with high capacity Reedwood cartridges (10 and 50 Gigabytes compressed). The objects archived by this service are files. There are several different layers of service with regard to file retrieval time and file duplication. The STAF is in charge of all data migration involved when changing from old to new media or to a new technology medium. They do not affect the upper layers of the system.

Formats. The format of data stored must be independent of all operating systems. In practice, experiment data is usually in IEEE or ASCII code and divided up into sequential files. The application of CCSDS encoding for times and dates is compulsory for all record structure files. The syntax and semantics of each file must be described with EAST and a DED unless self-descriptive structures such as FITS or NCAR are used. As far as documentary information is concerned, no reference standard for the internal representation of documents has yet been applied.

Data Management

Data management revolves around use of a graph describing data collections and objects. For the purposes of simplification, this graph is usually known as a data graph. It is oriented and non-cyclical. The relations associating a node with its descending nodes are (from an object-oriented viewpoint) inheritance and composition relations. A data set, also known as a terminal collection, thus inherits the characteristics of all the collections above it.

Documentary information, browse data and event tables are also managed through graphs which are nonetheless distinct from the data graph. The graphs contain either explicit metadata or references to external files or documents.

1.4 ACCESS

Access facilities are seen by the user through a WWW server. These facilities include aids to search for data collections and objects, means of retrieving certain metadata (such as documents and catalogues) immediately, ways of ordering data products which include special protective mechanisms for data not made public, and finally, generation and delivery of these data products.

Finding Aids

The aids to search data of interest to the user are based on navigation within the different graphs: the experiment data collection and object graph, the browse data graph, the documentary object graph and the events table graph. These graphs are independent but a certain number of links are used to move from one to another. Navigation within the graphs is, depending on the case, through criteria such as a keyword (parameter measured, location of measurements, etc.), time or other types of criteria.

The data object and collection graph grants several views of the data, and the final objects may be selected after several navigations within the graph.

The events table graph may be used to make indirect selections over time, such as selecting only data corresponding to a given instrument operating mode, or data corresponding to the periods during which a particular type of magnetospheric event was observed, etc.

These aids may be used to select data which is stored either on the main archive site (at CNES) or at an associated laboratory.

Security

Without exception, metadata is visible and accessible to the general public without any prior authentication. On the other hand, data may only be ordered by a user previously authorized by the CDPP, as it normally implies the consumption of resources. The user makes his request for authorization by a form available on-line, indicating his name, e-mail address, the name of the laboratory he belongs to and the reasons for his request. Once the user has received authorization to order products, he must authenticate his request (name and password) before ordering.

Data archived by the CDPP is usually public in nature, but in the case of recent data, data ordering may be temporarily restricted to one particular user group. The system must therefore be capable of handling access rights to the service (for ordering data) independently from access rights to the data itself.

Finally, the system is designed and has a number of protective measures such that any accidental or deliberate modifications to the data stored in the Center may be avoided.

Customer Service/Support

The system can handle profiles peculiar to each user, taking into account in particular the capability of the network linking him to Internet and the laboratory to which he belongs (laboratories directly supported by CNES, laboratories involved in cooperative projects with French laboratories, other laboratories, etc.).

The CDPP has a customer support team able to reply to technical questions (how to use the system, read data, etc.). This team can also direct the users to the Principal Investigator or data producers.

Data Transformation before DIP delivery

The data objects distributed to scientific users are not necessarily identical to the data objects stored in the system. Depending on the standards respected and tools available, a certain number of transformations of archived objects may be requested, in particular:

– Time-related retrieval which provides data corresponding to one (or more) time periods specified by the user. This kind of retrieval is only possible when times and dates have been encoded in compliance with CCSDS recommendations.

– Retrieval of fields, which permits the user to select fields of interest on the basis of an EAST data descriptor.

These transformations are known as ‘subsetting services’. Other such transformations are planned for future versions, so as (for example) to be able to deliver data in the user's native machine format, or deliver data as physical values although it is stored as raw values.

Media/Network Use for DIP deliveries

The data from Plasma Physics experiments is often bulky (a data set often contains between ten and several hundred Gigabytes). It is not planned to systematically create pre-defined, widely disseminated products as is often the case for planetary data, particularly as users are often interested in a specific period of time and not the whole data set.

Products may be delivered either over a network or on a variety of media (currently CD-ROM, DAT or Exabytes). The choice between these two types of delivery depends on the capacity of the network between the user and the CDPP at any given time.

As far as network deliveries are concerned, the system proposes the HTTP protocol at the user's initiative or the FTP protocol at the CDPP's initiative, but at a time specified by the user. The latter choice is subject to certain constraints. Deliveries of data via a network offer optional data compression and grouping facilities in the form of .tar files.

Pricing Policy. The pricing policy has not yet been fully determined. It will include an invoice for dissemination of data on an external medium (CD-ROM, DAT, Exabytes).