The MAINSOURCE Architecture
Object-Oriented Database
The MAINSOURCE architecture implements a proprietary object database at its core. Object Oriented technology is a relatively new and powerful tool for IT professionals but has a reputation of being slow and cumbersome. MAINSOURCE has bridged the gap by meta-tagging complex data types with a powerful 20 character meta-tag which is stored in any relational database. This provides for rapid retrieval of information (actual testing has shown certain configurations can support greater than 1 million transactions per second).
Universal Data Type
A foundation goal of the architecture of this product suite was the ability to handle data of ANY format that exists today and in fact be 'data type agnostic'. The data type is therefore irrelevant in the architecture so that new data types, yet to be defined or created, will be automatically accommodated.
Object-level Security
Security is a major consideration for a Production-Class system. Our architecture foresaw the need to provide sophisticated yet simple to implement object-level security. Not only do we implement a robust "role-based" or "Policy-based" security structure, we also allow any object with a collection of many terabytes of objects to carry its own security profile.
To accommodate highly secure applications, the meta-data tag also has a flag signifying 'external security' providing the 'hook' for applications that require implementations to comply with their existing corporate security infrastructure.
Object Replication
A major feature of the MAINSOURCE architecture is the ability to implement object replication invisibly and transparently to existing systems. Many of today's major infrastructure solutions have a 'big switch' approach where at one point in the implementation IT is forced to throw the big switch to move to the new system. The MAINSOURCE meta-tags can be added on-line and in-service through standard operations. The raw information ( video, audio, images, text, etc.) can remain in their original form, in their original databases and on their original servers. However, MAINSOURCE's versioning and rendition capabilities can be utilized to maintain transparently, a corporate or centralized copy of any objects. This is also useful for replicating highly used data geographically closer to the viewers. Object replication also ensures data integrity by virtue of the object architecture.
Scalable Architecture
From the original blue prints, MAINSOURCE's architecture was built to handle virtually unlimited data storage. A virtually unlimited number of Data servers can be distributed to provide massive scalability. The architecture provides the ability to reference 4 billion information servers, each of which is capable of identifying trillions of data objects. This distributed multi-server architecture also means that applications can be scaled to service millions of simultaneous users in a high-performance, high-availability configuration.
Indigenous Storage
Indigenous storage refers to the architectural approach that focuses on tagging data 'where it lives' and not demanding that the data be moved or copied into a central database to provide ubiquitous access. Providing broad access should not require moving data to a central site, but can be replicated according to business rules.
Universal Tagging
Universal tagging refers to the ability of the architecture to tag and store and object regardless of data type, technology or location of creation.
Universal Auditing
The MAINSOURCE architecture allows for auditing to the finest granularity possible, the transaction level. Every type of access can be audited. This is essential in some applications such as Healthcare where access to patient records must not only be controlled but also auditable. It is one thing to say "Only ICU nurses have access to this patient's record" and quite another to say, "Here is a list of the people who accessed this record and the transaction record of what they did with the information."
Portable Kernel
To make the architecture as transportable as possible, the code has been implemented using ANSI standard C, widely recognized as one of the most transportable production-class computer languages. Initially launched on OpenVMS, the code has been recompiled on other platforms with little effort.
Integral TPS
Any production class system that must be massively scalable must be built on a transaction-processing framework. MAINSOURCE integrates state of the art commercial transaction processing systems tuned for any platform.
Asynchronous Architecture
A key design consideration of the architecture was the ability to maximize performance. Many systems lose any performance advantage due to their serial processing nature. The MAINSOURCE architecture allows for parallel searching and retrieving. An application can be requesting a metatag for a piece of key information while at the same time receiving the data stream from previous requests.
High Performance
Due to several factors such as state of the art disk systems (RAID, Juke boxes), asynchronous architecture, separating the metadata from the actual data, and high performance transaction processing systems, the MAINSOURCE suite provides outstanding performance on the high end. Actual testing in high-end configurations has shown transaction rates of over 1 million transactions per second for metatag queries.
Rapid Universal Application Deployment
Once data is ubiquitously available through meta-tagging, MAINSOURCE's API allows for rapid deployment of new applications that span the data. Constructs of Folder, Document and Page are pre-installed and ample sample code using the API is also available for rapid deployment.
