Distributed Data Framework Executive Summary: Core Concepts and Introduction

Distributed Data Framework (DDF) is a free and open-source common data layer that abstracts services and business logic from underlying data structures to enable rapid integration of new data sources.

Licensed under LGPL , DDF is an interoperability platform that provides secure and scalable discovery and retrieval from a wide array of disparate sources.

DDF is:

DDF is comprised of several modular applications, to be installed or uninstalled as needed.

The following conventions are used within this documentation:

Questions about DDF should be posted to the ddf-users forum or ddf-developers forum , where they will be responded to quickly by a member of the DDF team.

DDF provides the capability to search the Catalog for metadata. There are a number of different types of searches that can be performed on the Catalog, and these searches are accessed using one of several interfaces. This section provides a very high-level overview of introductory concepts of searching with DDF. These concepts are expanded upon in later sections.

There are four basic types of metadata search. Additionally, any of the types can be combined to create a compound search.

In DDF, resources are the files, reports, or documents of interest to users of the system.

Metadata is information about those resources, organized into a schema to make search possible. The Catalog stores this metadata and allows access to it. Metacards are single instances of metadata, representing a single resource, in the Catalog. Metacards follow one of several schemas to ensure reliable, accurate, and complete metadata. Essentially, Metacards function as containers of metadata.

Ingest is the process of bringing data resources, metadata, or both into the catalog to enable search, sharing, and discovery. Ingested files are transformed into a neutral format that can be searched against as well as migrated to other formats and systems. See Ingesting Data for the various methods of ingesting data.

Upon ingest, a transformer will read the metadata from the ingested file and populate the fields of a metacard. Exactly how this is accomplished depends on the origin of the data, but most fields (except id) are imported directly.

The Catalog Framework can interface with storage providers to provide storage of resources to specific types of storage, e.g., file system, relational database, XML database. A default file system implementation is provided by default.

Storage providers act as a proxy between the Catalog Framework and the mechanism storing the content. Storage providers expose the storage mechanism to the Catalog Framework. Storage plugins provide pluggable functionality that can be executed either immediately before or immediately after content has been stored or updated.

Storage providers provide the capability to the Catalog Framework to create, read, update, and delete resources in the content repository.

See Data Management for more information on specific file types supported by DDF.

The Catalog Framework wires all the Catalog components together.

It is responsible for routing Catalog requests and responses to the appropriate source, destination, federated system, etc. 

Endpoints send Catalog requests to the Catalog Framework. The Catalog Framework then invokes Catalog Plugins, Transformers, and Resource Components as needed before sending requests to the intended destination, such as one or more Sources. 

The Catalog Framework decouples clients from service implementations and provides integration points for Catalog Plugins and convenience methods for Endpoint developers.

Federation is the ability of the DDF to query other data sources, including other DDFs. By default, the DDF is able to federate using OpenSearch and CSW protocols. The minimum configuration necessary to configure those federations is a query address.

Federation enables constructing dynamic networks of data sources that can be queried individually or aggregated into specific configuration to enable a wider range of accessibility for data and data resources.

Federation provides the capability to extend the DDF enterprise to include Remote Sources, which may include other instances of DDF.  The Catalog handles all aspects of federated queries as they are sent to the Catalog Provider and Remote Sources, as they are processed, and as the query results are returned. Queries can be scoped to include only the local Catalog Provider (and any Connected Sources), only specific Federated Sources, or the entire enterprise (which includes all local and Remote Sources). If the query is federated, the Catalog Framework passes the query to a Federation Strategy, which is responsible for querying each federated source that is specified. The Catalog Framework is also responsible for receiving the query results from each federated source and returning them to the client in the order specified by the particular federation strategy used. After the federation strategy handles the results, the Catalog returns them to the client through the Endpoint. Query results are returned from a federated query as a list of metacards. The source ID in each metacard identifies the Source from which the metacard originated.

DDF can be configured to receive notifications whenever metadata is created, updated, or deleted in any federated sources. Creations, updates, and deletions are collectively called Events, and the process of registering to receive them is called Subscription.

The behavior of these subscriptions is consistent, but the method of configuring them is specific to the Endpoint used.

Endpoints expose the Catalog Framework to clients using protocols and formats that the clients understand.

Endpoint interface formats encompass a variety of protocols, including (but not limited to):

The endpoint may transform a client request into a compatible Catalog format and then transform the response into a compatible client format. Endpoints may use Transformers to perform these transformations. This allows an endpoint to interact with Source(s) that have different interfaces. For example, an OpenSearch Endpoint can send a query to the Catalog Framework, which could then query a federated source that has no OpenSearch interface.

Endpoints are meant to be the only client-accessible components in the Catalog.

DDF can be made highly available. In this context, High Availability is defined as the ability for DDF to be continuously operational with very little down time.

In a Highly Available Cluster, DDF has failover capabilities when a DDF node fails.

This setup consists of a SolrCloud instance, 2 DDF nodes connected to that SolrCloud, and a failover proxy that sits in front of those 2 nodes. One of the DDF nodes will be arbitrarily chosen to be the active node, and the other will be the "hot standby" node. It is called a "hot standby" node because it is ready to receive traffic even though it’s not currently receiving any. The failover proxy will route all traffic to the active node. If the active node fails for any reason, the standby node will become active and the failover proxy will route all traffic to the new active node. See the below diagrams for more detail.

There are special procedures for initial setup and configuration of a highly available DDF. See High Availability Initial Setup and High Availability Configuration for those procedures.

DDF incorporates support for many common Service, Metadata, and Security standards, as well as many common Data Formats.