Planning a project¶
The main planning task before starting a Sesam project is to create an overview of what master data types are needed, their sources, and how they should be organized inside Sesam.
The dataflow can be documented by filling this
template by following these 4 steps:
Identify what master data you need and where this data must be available
Identify where the master data exists today
Identify the system-independent globals (common data grouping) needed to store the master data in Sesam
Connect the systems and master data types to the globals
Identify master data¶
Based on the functional needs of the project, identify the data types at a high level, and where this data must be available.
For the master data giving context to the data platform, driving analytics, AI, web and Apps, select the appropriate data platform components from your preferred vendors. E.g., Azure, Google etc.
For master data needed to synchronize with new or existing business systems, identify their API, data protocol, the endpoint URI, and what connector can be used to communicate with them.
Identify master data sources¶
Identify the source systems that contain the necessary data to fill the master data requirements of the project.
Identify their API, data protocol, and what connector can be used to communicate with them. Decide if the system has test and prod environments and find out what their endpoint URIs are.
To be able to effectively implement the project, a system owner and preferably a data owner for each system should be identified.
Once all sources have been identified, an initial priority of the different sources can be set. Central objects will always be stored in multiple systems, and prioritizing what system will override others is important to support multi-master data synchronization.
Identify globals¶
Once the necessary master data sources and targets have been found, the source master data must be grouped into globals in Sesam.
Group data by type, not role. One object should never fit in more than one group. Start generically in this phase. Re-grouping to more detailed groups can be done with time.
Samples of a good way of grouping into globals:
Person - Don’t use employee, contact person, or user
Business – Don’t use supplier, customer, or partner
Product – Don’t use screw, nut or bolt
Connect data types to the globals¶
Finally connect all identified data types to their respective globals to complete the data flow architecture.
Incoming data may be split into different globals, and outgoing data may need to combine several globals.