ALFA is an easy-to-use data modelling platform. It is agnostic of how the model will be implemented - i.e. it is not coupled to a particular programming language, technology or framework. It focuses on letting Modellers express their abstractions and rules concisely close as possible to how they best see fit. Those models can be extended by Architects with Governance and controls. Engineers can then build applications using ALFA models and generated code.

What’s in it for you?

Decision Maker

  • IT teams working with well-defined models are able to maintain delivery-velocity
  • Changes to models are easily accommodated
  • IT focus on high level, value-added capabilities

Data Owner

  • Consistent data format and naming, easily understood by all
  • Eliminate need for multiple reconciliations and manual adjustments to focus on business value activities
  • Easily see where data comes from and where it goes to
  • Be in control of data at all times, including versioning, knows what data resides where, published and consumed
  • Always live, navigable, versioned documentation

IT Architect

  • Centralized, federated or hybrid data models ownership
  • All models, rules and metadata visible to stakeholders to facilitate agile development practices
  • Data-Product driven model sharing between Data Producers and Consumers
  • Single source of truth. Data lineage, Capture ownership, and all metadata alongside the model
  • Support on-premise or multiple cloud solutions
  • Extend governance by attaching PII, masking tags to identify and apply compliance rules


  • Faster application development times
  • Fewer re-writes or code change due to ambiguous specifications
  • Promotes code reuse
  • Avoid mundane data-related tasks, deliver business applications and value
  • Using clean data, look at value-added tasks - data analytics, enable AI/ML

An Example ALFA model

ALFA data models and rules are written in text files. The ALFA compiler validates the model, and generates code for the supported programming languages or frameworks, along with documentation of the model. This can be an iterative process, bringing together modellers, developers, even users. ALFA becomes the common language to describe data regardless of technology or programming language.

ALFA models with data quality rules can be used to validate data in a variety of formats.

An detailed example showcasing number of ALFA modelling features is shown below:

namespace sample

// A trait is a set of reusable attributes
trait Person {
   FirstName : string
   LastName : string
   DateOfBirth : date

enum ServiceLevelType {
   Standard Gold Platinum

key EmployeeKey {
   EmpId : uuid

// Entities have keys and can include pre-defined traits
entity Employee key EmployeeKey includes Person {
   HiredDate : date

// Entity keys can be predefined or self-defined
entity Customer key ( Id : uuid ) includes Person {
   Email : string
   Accounts : set< string >
   DeclaredIncomes : list< double >
   RelationshipManager : EmployeeKey?

   // This is a calculated field
   ServiceLevel : ServiceLevelType = CustomerLib::serviceLevel( DeclaredIncomes )

   // To validate, ensure customer is an adult.
   // Multiple asserts can be defined for a user defined type and gets used to enforce Data Quality.
   assert IsAdult {
     let age = dateDiff( today(), DateOfBirth ) / 365
     if ( age < 18 )
          raise error( "Customer ${FirstName} ${LastName} aged ${age} is not an adult" )

// Complete and reusable logic rules can be expressed in ALFA. These get generated into various target languages
library CustomerLib {
   // The serviceLevel is calculated by adding all values and assigning a level based on total income range
   serviceLevel( vals : list< double > ) : ServiceLevelType {
       let total : double = reduce( vals, 0.0, (acc, i) => acc + i )
       return if   ( total < 100000 ) ServiceLevelType.Standard
              else if ( total < 1000000 ) ServiceLevelType.Gold
              else ServiceLevelType.Platinum

// This is a service/API defined in ALFA. This can be generated into a language API, REST endpoint etc.
service CustomerSvc {
   getCustomerByEmail( email : string ) : Customer?

Test drive a live instance of ALFA web-based tools using the example above -


A summary of the main features is listed below.

  1. Domain specific language to express data models with minimalist and clear syntax.
  2. Ability to create fully documented strongly typed data models.
  3. Compiler that will perform extensive compile-time static analysis, and accurately report all errors, if any.
  4. Assert data validation rules to enforce DQM (Data Quality Management) in applications using ALFA.
  5. Extensible type definition system to allow composing derived types including templated types.
  6. Ability to express complex default values for fields of any type, including user-defined types and collections.
  7. Describe services (APIs) using modelled objects.
  8. Provision to define versions of service and type definitions.
  9. Code generators to create programming models/APIs from the ALFA data model.
  10. Development, packaging and build tools to be used as part of a model driven, iterative, rapid development process.

ALFA Files

ALFA has 3 type of file artifacts - model files, project files, module files.

  1. ALFA model files are identified by their file extension .alfa. ALFA files can contain one or many model definitions.
  2. ALFA project files are created with the .alfa-proj.yaml or .alfa-proj.json extension. Project files contain settings for the compiler and dependencies to other modules.
  3. Compiled ALFA files and a project file can be used to create a ‘module’. A module is created as a file.

ALFA Generated Code

Once a model is successfully compiled, it can be used to generate code for a target language or framework. For example a Java POJO model, which depends on an ALFA Java runtime library, containing various utility classes such as JSON serialisers.

In addition to generated code, ALFA generates model documentation in HTML using embedded comments as documentation with navigable dependencies and links to usages. The documentation forms an integral part of the modeller’s toolset to share ideas and knowledge.

The ALFA compiler and toolset can be incorporated into a continuous delivery pipeline where code and documentation generation, and upload of artifacts to repositories can happen as part of the pipeline. Any breaking changes are caught at compile time before the code gets promoted.

ALFA In Practice

  • Truly Continuous, Model-Driven-Development
  • Fits in with your working practices
    • ALFA does not mandate a particular Software Development Lifecycle (SLDC) or development principles.
    • Users can easily incorporate ALFA into their existing SLDC and Data Model Governance processes and guidelines.
    • Works with your existing Source Control, Repositories, Build tools etc.

The diagram below shows a typical usage scenario of an organisation that has integrated ALFA into their SDLC.