The Technology Architecture of an Information Delivery Solution (IDS), typically consists of a number of logical nodes: The Sourcing Node, The Storage Node, The Computational Node, The Aggregation Node, The Visualization Node and lastly Dissemination Node. These logical nodes each have very specific functions and capabilities within the entire chain of processes.
The most important technical aspects are as follows: Each node can be made up of one or more hardware component (For instance for the Sourcing Node, there may be 1 or 3 or 6 or n number of physical Staging Servers). The total system is made as parallel as possible so that one ends up with a stream of processes in which all may work simultaneously and in tandem. Hardware and Software resources are allocated to a particular node in order to balance the performance by increasing throughput and by removing bottlenecks. This ensures that the system is massively scalable, because processes are decoupled and running asynchronously. At the core of the Datastore remains a RDBMS but, recently with Oracles two layered database approach, inserts are done at the lower layer while the instant aggregation allows querying from the higher layer. In spite of this, the ability of the database management system itself is to perform operations in parallel and to optimize data retrievals remain a core foundation of a good IDS. (This removes the need for separate OLAP engines).
The computational nodes are of paramount importance for Information creation, as many Information Analytical programs require both numerous repetitive iterations through the database structures and data sets, as well as the derivation and computation of generated simulative data - such as in market forecasting, actuarial simulations, recalculations and restatement of previous financial statements, etc. For this purpose, the computation nodes are implemented by means of a set of specialized GPU cards, (each have thousands of CPU Cores, and each having the ability to perform a vast array of mathematical and statistical functions on large sets. Not only is the number of CPU cores important, but the ability to process Single Instruction on Multiple Data Cells (SIMD) as well as Multiple Instruction on Multiple Data Cells (MIMD). This is of paramount importance to increase computational band width. Recently the Aggregation Node has also received special attention from GPU technology, rather than having ROLAP, MOLAP, HOLAP type of arrangements in and around the dataset. The data of interest is loaded in memory from where literally thousands of CPU cores are thrown at the problem. Since the principle of divide and conquer works so well, GPU's are experts of working in this manner. The results is near instant performance, and to allow us to develop applications in another manner. The traversal paths are followed as the information gets navigated on a just-in-time basis and there is no point pre-processing. Speed is no longer the obstacle, meaning that the visualizations and presentations are live and vivid.
Interestingly, the Visualization Node Capability often relies on the processing and rendering capability of the Delivery Devices. Graphs displayed as an analysis surface resembling the waves of the sea, make the graph highly visual, dynamic, relevant, and the swift movements immediately alerts the viewer to some major events that have taken place. Colour, change of colour, and movements are very important for Visualization. For instance, when data is codified appropriately and then plotted, the representation reveals patterns in graphs and colour previously unimaginable. The Information User quickly associates this with a meaning related to the business thus the insight and knowledge is rapidly attained.
Practical challenges remain with the dissemination of Information to so many end-users on the devices of choice. Fortunately, smart device makes it much easier to deploy to the correct audience, but managing, storing and using the sheer amount of data remains.
Our deep study of technology keeps us ahead. Years ago, when India was behind with respect to economy and technology, they set a view to become the nation providing the most affordable technologies and resources to the world. This vision dramatically changed their position in the work place.
Lured by the sheer cost-benefit value of this arrangement, spread the use of Indian resources like wild fire. This outcome of course brought many other problems and so today we understand that quality and technological excellence are crucial for sustained market dominance - as is the case with Apple devices.
Infomet brings business and software engineering technologies to the fore because without it the technology will simply not have the longevity it needs.
No longer are we living in a world where monolithic computer systems weigh tons, stand rooms full and cost gazillions of dollars, prevails. The world has changed and we need to focus on the top 20 information technologies of our time.
Mobiles are not portable, people are transportable. But we need small computers and intelligent devices everywhere. This implies cloud based computing with accessible devices everywhere.
Focus on the relevant technologies. All else is a waste of time. Think carefully about which technologies will prevail in future. Some are just fads - they will fade away and so will your effort and investment. Lastly, engineer your systems properly using the Modelling Tools and Specification Repositories. Often our quickest way to migrate away from obsolete technologies, is to simply re-generate new products from quality meta data, using new patterns and templates. Invest in tools.
What excites me is the opportunity to engineer
World Class Infostacks (Infomet's Information Delivery Solutions). The Hardware, Operating Systems, Network, Specialised components as well as Data Technologies and Software.
The more I look at what is on offer by major IT Suppliers today, the more I marvel at what we achieved at Infomet. These systems are today successfully implemented at major banking institutions.
Infomet © 1986,1995,1997,2001,2004,2009,2011,2015 : Intellectual Property Policy