I3S Final meeting

Session 3

VTL Tools

What is VTL?

The Validation and Transformation Language (VTL) allows the formal expression of treatments to:

  • validate statistical data
  • calculate derived data

What is VTL?

  • an SDMX standard
  • grammar is formally defined (EBNF)
  • strong support by Eurostat, the ECB and others

What is VTL?

VTL is user oriented and provides a technology-neutral and standard view of statistical processes at the business level

What tools for VTL?

  • VTL is a language, so it has to be edited (-> editor)
  • VTL expressions or rules can be managed and shared (-> rule manager)
  • VTL is technology-neutral, so has to be translated to an executable language (-> execution engines)

EBNF helps hugely to build tools for VTL thanks to Antlr

What was done in I3S?

What was done in I3S?

Context

  • Eurostat developing a rule manager and editor
  • SSB had a Java execution engine for VTL 1.1
  • Insee developing JavaScript engine and editor

What was done in I3S?

Achievements

  • Eurostat-Insee meet for technical alignment of editors
  • Eurostat open-sources its VTL editor
  • Insee drops own editor & contributes to Eurostat's
  • Insee develops Trevas with SSB contribution
  • Trevas added to the I3S deliverables

Zoom on Trevas

Zoom on Trevas

  • a Java engine for VTL 2.0
  • provides an abstract definition of a Java VTL engine
  • also two concrete implementations
    • in-memory engine for relatively small data
    • Apache Spark engine for Big Data production environments

Zoom on Trevas

  • Work in progress (coverage improving)
  • Open source (MIT) on GitHub
  • Published on Maven

Zoom on Trevas

Demo

(long version available on video)