acs-reference Design Documentation
I. Introduction
Reference data services are often overlooked in the rush to get
coding. Much of the code is redundant or of similarly patterned
implementations. This package intends to address some common features
and needs.
II. Historical Considerations
Before the existence of acs-reference, the ACS required that you preload some tables in a script to get some basic reference functionality. There were many problems with this:
- No easy way to find out what reference data even existed.
- No way to find out how old the data was.
- No way to find out where that data came from.
- Very US/English slant on the data.
III. Design Tradeoffs
Primary Goals
- This system was designed with maintainability and reusability as its primary goals. By wrapping a layer around all of the reference tables we have increased the maintainability immensely.
- Another goal was to bring together many different types of data and present them in a logical fashion. It was amazing how little of this data is available on the internet in a database friendly form.
Performance
When updating the reference tables their is overhead due to the fact that the table is registered with the repository. This should rarely occur anyway as the tables are only added once.
By not having the actual data itself in the acs-object system, subsequent additions and deletions to the reference tables themselves are unaffected by this overhead.
IV. API
See api-browser
V. Data Model Discussion
The UNSPSC reference data has a data model for handling data
revisions. An application can determine any new/revised category based
on existing, obsolete data.
VI. User Interface
Their is no end user interface.
There needs to be some kind of admin UI to report status and possibly
manage updates per requirements.
VII. Configuration/Parameters
None
VIII. Future Improvements/Areas of Likely Change
A server based update mechanism will be supported. This will allow for tables to be updated (and preferably diffed) instead of being reloaded with a package upgrade.
An interface to produce xml/csv from the reference data would be a nice service to the community (allowing legacy applications a way to import this data).
IX. Authors
Jon Griffin