To build a truly revolutionary data integration platform, we embraced specific, no-compromise, design and architecture maxims.
Process in parallel never series
Follow SOA and loosely coupled design
Employ massively parallel processing (MPP)
Use clustering to scale on standard hardware
Innovate fast and flexible data-structures
Be platform/OS agnostic using open standards
Leverage the cloud and virtualization
Employ smart-caching and data compression
Short explanation of how DataRoket works
DataRoket uniquely leverages multi-threading, massively parallel processing, a flexible and fast columnar/relational data model, RIA, cloud delivery and virtualization. DataRoket delivers an entirely new breed of data integration and performance without compromise.
Connect Disparate Data—Anywhere— Now
Ideally, all important data would sit in central databases, with common nomenclature. In the real world, data sits in databases, ERP systems, vendor reports, web-services, various date performance accelerators and reporting tools. DataRoket solves the disparate data challenge: our data connectors read databases, systems (SAP BW, Business Objects Universes) with a robust SOA structure, enabling a new proprietary connector to take a few weeks to build— not months.
Best of Both Worlds - Hybrid Columnar and Relational Row Pointer Model
Although columnar databases querying and analytics provide performance benefits, significant limitations remain, making a basic columnar data structure unsuitable for a granular data. The ability to do updates and deletes of records in a pure columnar format is almost impossible. Vendors using columnar data structures state "a good rule of thumb-- fewer than 1% of the total SQL statements should be Deletes or Updates." The result: differential measures and near real-time data updates are simply not possible in the columnar model.DataRoket solves these problems.
Our data model leverages the advantages of the column while maintaining row pointers (persisted and in-memory) to enable incremental and real-time asynchronous updates and deletes-- without always having to rebuild the data set.
Massively Parallel Processing (MPP)
Today’s legacy applications and data structures often fail to scale and meet today’s performance needs. DataRoket leverages MPP to use off the shelf hardware and scale by simply adding another machine with no limit.
Full Service-oriented Architecture (SOA) and Open Standards
DataRoket meets all SOA requirements by allowing each module to run on a separate application server and operate as a full application. This enables the deployment of the modules in environments that can support different loads without spending resources on supporting unnecessary aspects of the application. DataRoket passes information and process during published web services, enabling a level of straight-forward integration unavailable with other applications. Because of the open standards coupled with the SOA architecture, DataRoket acts as an unparalleled data integration application. DataRoket uses standard formats such as XML, published web services and open standards across the interfaces. This enables straight-forward integration across a variety of options.
DataRoket was built from the ground up to accommodate cloud computing. DataRoket's minimal footprint means that it can be offered via the cloud or data center deployments. By leveraging SOA, DataRoket modules can be segmented, giving customers the choice of which aspects remain in the data center and which can be pushed to the Cloud. Through DataRoket’s virtualization, utilization of less hardware, and sharing resources in the cloud, we’ll decrease your energy consumption, equipment disposal, and drastically decrease your Total Cost of Ownership (TCO).