What is DObjects?
DObjects is an easy-to-use and scalable distributed data objects framework for heterogeneous systems being developed at Emory University. The framework has a number of unique features and contributions that distinguish it from current solutions.
First, the system employs a novel distributed and decentralized architecture. It uses a metacomputing platform as a resource sharing substrate. The system consists of independent nodes which form a virtual abstraction of a single database management system. Each node can either be responsible for retrieving data from data sources through data adapters or simply provide computational power which is used by others. The system scales well when only few or thousands of nodes are used.
Second, the system provides convenient and transparent querying mechanisms for data objects. Front-end users build queries and get back an array of data objects called persistent entities as the response to the queries. DObjects' query language strictly follows object-oriented fashion of data representation. When query is created, users have a possibility of constraining results in terms of populated attributes (the system fills in only desired attributes in result persistent entities) and conditions that have to be satisfied by returned data objects. Data from multiple heterogeneous sources are joined transparently from user's point of view. Each persistent entity is a wrapper for data obtained from data source.
Third, the system implements a distributed query optimization scheme based on a cost space model that optimizes both throughput and response time of the system. Every time a query is executed, it is deployed on some nodes of the system. The deployment process is driven by various factors, such as desired data placement, current load of particular nodes and distance between nodes in terms of network latency.
Finally, the system is a general purpose distributed data objects framework. It supports various options of query types, such as ad-hoc queries, batch queries and continuous queries for various applications.

The figure above presents our vision of deployed DObjects framework. The system has no centralized services and thus allows system administrators to avoid all the burden in this area. It uses the metacomputing paradigm as a resource sharing substrate. Each node in the system provides its computational power that can be used by others during query execution. In addition, nodes can run data adapters which pull data from external data sources and transform it to uniform format that is expected while building query responses. Front-end users can connect to any system node; however, while the physical connection is established between a client and one of the system nodes, the logical connection is established between a client node and a virtual database system consisting of all the participating nodes.

What do I need to know to use the system?
In general, even though DObjects is a distributed framework, you do not need to be an expert in distributed systems to use it. It has simple end-user interface which is expressed as classical database interface that can be queried. The only difference is that, you have to use query language which the system provides (not SQL, but more object-oriented language).

What are the benefits of using this framework?
DObjects is framework that facilitates integration of data from multiple heterogeneous sources. By defining system configuration you can easily provide information how data from different sources should be mapped to global, uniform schema. After this step, you can start system and query for data. Even though your connection happens to be between client and one of system nodes, you can use whole virtual database system, e.g. you can use computational power of all the nodes or query and operate on data from any datasouce uniformly.

System requirements
DObjects is written on top of H2O Java metacomputing framework, so it is platform independent. It was tested on Windows, Linux and SumOS and was running well. You have to just remember about appropriate firewall settings which will enable interactions between distributed components.