UDM Appendix C - Governance

This section describes the principles that guide the governance choices made for the Ed-Fi Unifying Data Model (UDM).

Influences

Changes to the data model are influenced by a number of sources:

  • Field Implementation Reports and Artifacts.
  • The Ed-Fi Community and governance groups and processes
  • Ed-Fi Special Interest Groups.
  • Data Standard RFC Comments.
  • Bug Reports (e.g., in Ed-Fi Tracker).
  • Common Education Data Standards (CEDS) and other standards related to K–12 education.
  • Industry and International Standards, where established and appropriate for use in K–12 education.

These influences combine to define each subsequent version of the Ed-Fi UDM.

Making Changes

All changes to the Ed-Fi UDM have some impact and certain kinds of changes are easier than others. This section defines key points on how the Alliance thinks about changes to the Ed-Fi UDM. 

Principles

There are a few major factors that influence if a change should be made to the UDM.

  1. There is clear evidence of need across many education organizations within the Ed-Fi community. The Ed-Fi data standard exists to serve common and shared  use cases, and non individualized use cases. To ensure that changes serve the community, there must be evidence that the exchange of the elements is valuable to a reasonable segment of the community. There is no magic way to determine how much evidence is needed; for a minor addition it might be enough to have 2 or 3 examples of field need, and for larger ones, we must be confident that the need is greater.
  2. There must be sufficient evidence from which to design a data model. Even if we have established the need, we must still as a community have access to sufficiently detailed use cases and other artifacts (such as source system data models) to understand the right data model. Aligning on a sub-optimal data model can be very painful, as, once operationalized, a standard is very hard to change. For this reason, considerable effort goes into this task.
  3. We must consider if the source systems from which the data is flowing is part of the Ed-Fi community and/or conversation. Since the purpose of the Data Standard is to standardize data exchange and to do so in a community-centric fashion, it is important to engage providers and others who manage systems from which the data originates. Efforts to grow the standards should be actively seeking to incorporate those voices.

Additions

  • Adding entire domains or major entities to the Ed-Fi UDM generally require a strong pull from the Ed-Fi Community or a significant technology change in the education data field.
  • Adding domains or major entities also generally happens in conjunction with field work such as proof-of-concept implementations, or deconstruction of technology submitted to the Ed-Fi Exchange.
  • Adding elements to an existing entity requires a lower burden of proof, but still requires documentation of specific use cases from field work.
  • The Alliance periodically reviews how field work is extending concrete implementations. In cases where field work has extended the standard in the same way, the Alliance will work with the community to propose and add Ed-Fi UDM with elements to support those extensions natively. This practice helps reduce ecosystem fragmentation and evolve the scope of K–12 interoperability.
  • Adding required additional elements is often a breaking change — so the Alliance generally waits until a major release to do so. Optional fields may be introduced in a minor release if it appears it can be done without introducing major difficulty for current exchanges.
  • With every release, the Alliance reviews the then-current version of CEDS (primarily the entities and elements in the K–12 domain, but the entire model is scanned). This generates elements proposed for addition. Other data standards relevant to use cases supported by Ed-Fi are also considered and reviewed in a similar fashion.
  • Relatively minor changes (e.g., slight changes in definitions) or easily handled (and likely non-breaking) migrations (e.g., changes from an integer to a decimal), are generally easier and quicker to handle. Changes with a greater systemic impact (e.g., making a formerly required element optional) go through a more intensive review process.

Corrections

Corrections made to fix obvious flaws are (relatively) straightforward.

  • Corrections to annotations, definitions, or other non-technical artifacts with no impact to implementations are usually approved and implemented. For example, a spelling error in a entity annotation or general documentation will be made at the earliest opportunity.
  • Corrections and naming inconsistencies in field names are also usually approved. Since these are often breaking changes to implementations, we typically wait for a major release or for a point when other breaking changes are also being released.
  • Modeling errors are occasionally found, or weaknesses become apparent over time. If the error or weakness is in a domain not widely in use (operationalized), the change is usually made in a subsequent minor release. If the error would cause substantial ecosystem impacts, it may be held for a major release.

Deletions

Deletions can be problematic for implementers, so we tread carefully here. However, judiciously applied, deletions can keep the standard simple and bloat-free.

  • Every update cycle, the model is scanned for elements that appear to be unused in field work and they are flagged for removal.
  • Where possible, elements are deprecated before removal. This provides the community a transition period and also allows for implementers to alert the Alliance about the need met by the entity.
  • Where applicable, mappings or options for handling deleted or deprecated elements are provided. 

Transparency

As noted, the Alliance takes input from all sources at every opportunity. Most Ed-Fi Community members have full-time jobs, so we generally reach out only when we have major changes — but those who follow the Alliance's work have many opportunities to provide feedback. Here's our general approach to ensure that we telegraph all changes before they happen:

  • To the extent practical, all Data Standard discussion is documented in the Ed-Fi Tracker DATASTD project. Anyone is invited to participate in that discussion (see the Ed-Fi Technical Community Guidelines).
  • Every change is described in at least one public RFC before it's final. Most Data Standard releases that contain changes to the Ed-Fi UDM are preceded by multiple RFCs.
  • If scheduling permits, major changes are presented and discussed at the annual Ed-Fi Summit and/or the Ed-Fi Technical Congress, as is applicable in that release cycle.
  • The Alliance convenes an expert panel to reviews all domain additions, major entity additions, and major changes.
  • Community Governance structures also provide input mechanisms for specific domains or areas of the model.

So, if you care about this stuff, get involved. Comment and share ideas and feedback early and often. 

More Information

To find out more, or to get involved, see About the Ed-Fi Alliance, particularly the section on Data Standards