From a business point of view all IT systems can be divided into two major groups; Operational systems that support day to day business activities, and Analytical systems that. like a time machine, keep history of what happened in the past and as such provide a basis for making decisions what to do in the future, which is now days named as Business Inteligence or by it's acronym, BI. The ams is  initiated by our strong believe what Analytical component of BI should be and how it should be organised (although we may say that BI as a concept covers entire range of Company's activities and beyond, and that it has two areas of interests; operational and Analytical).
The most important element of an Analytical system is historical data. Operational data the very next moment become historical data. It is very important how the historical data is made available for analysis and where the data is located. Very often we come across opinion that the latest technology, BI packages, provide powerful tool for analysing distributed historical data.  

We think that the concentration of the historical data gives much better and richer potential.

We believe in Pyramid concept where all the hard work is done before the Data Mining tools come to the stage. This concept requires a bit more work which pays off in a long run.

The chart on the right shows the role of the ams in the BI story. 

AMS provides framework, modular structure, modular workflow and governance (monitoring and controlling) for all Application Building Blocks or ETL components, which allows building long term basis for the last phase in the system, performing complex analysis which includes Data Mining, Reporting, etc.

For Data Mining / Reporting component we recommend low or No cost Open Source Packages like MicroStrategy, Jasper, KNIME, Intelicus and similar.

In a nut shell a simplified Analytical system would have three components:

• collecting data
• storing data in a structure dictated by the next component
• using stored data.

 which, further translates into two major tasks for the ams;                 

• Extract
• Transform and
• Load data.

The most common method of organizing data is a Relational Database. We recommend no cost solutions like MySql and Postgres.

The chart bellow shows methods commonly used in physical implementation of the above concept.

The essence of an ETL is comprised of three functional elements. Extract, Transformation and Load. From Transformation point of view Extract is Input and the Load is Output.

The Input can come from two major types of sources:

• File of various types and formats,
• Data Base, again various types, but the most common would be a Relational DB

The Transformation has a capability to transform A to B according to predefined business rule.

The Transformation result can be Output into various output types and formats.

To all this we have to add a time dimension which would dictate the conditions and the frequency of the processing repetition.

Of course in reality the processing is often a lot more complex and hence it is a good practice to break it down to a smaller application building blocks.
The basic principals in structuring the ams application building bloks we have 4 grouping levels; task, procedure, phase and system.

Task is the only executable object in the  ams. It can be anything that can be executed on the platform. From a single command to a group of complex programs.
Task can have arguments. The ams allows task to be selected or unselected for execution.

Procedure is collection of Tasks placed in a sequence sequence which determines  order of Task execution.

Phase is collection of Procedures. Procedures could be mutually dependent. The condition to execute a procedure related to another procedure is defined by a dependency. Procedure can  be related, dependent on one or more  procedures or none. In the chart  on the right the procedure 4 depends on 2 and 3, and the procedure 1 has no dependency.

System is colelction holding one or more Phases.