Migrations
Pro feature
Parallel-run verifier
Run the incumbent and candidate pipelines over the same population and diff their canonical outputs field by field — the go / no-go evidence for a migration cutover.
How the parallel run works
What this is
During an SRS migration you run the old and new pipelines side by side over the same student population. The verifier aligns both canonical projections by (entity, id), compares every field, and scores agreement as a Data Health Percentage — 100% means the candidate pipeline reproduces the incumbent's output exactly.
How to use
- 1.Point projection A at the incumbent pipeline's canonical output and projection B at the candidate's.
- 2.Review the per-entity DHP — anything below 100% has at least one divergence to explain.
- 3.Triage the field-level diffs: mismatches are usually codeset or format drift; missing records are population alignment issues.
- 4.Re-run after each fix until the DHP is clean, then attach the report to the cutover sign-off.
Seeded demo — projections diverge deliberately so every diff status renders.
A · SITS canonical projection
↔B · Banner canonical projection
Overall DHP
88.9%
Entities
2
Records compared
6
Open diffs
5 (+2 missing)
Per-entity agreement
| Entity | Records | Missing in A | Missing in B | Fields compared | Matches | DHP |
|---|---|---|---|---|---|---|
| Student | 3 | 1 | 1 | 15 | 13 | 86.7% |
| Enrolment | 3 | 0 | 0 | 12 | 11 | 91.7% |
Field-level divergences
| Entity | Record id | Field | Status | Value in A | Value in B |
|---|---|---|---|---|---|
| Student | 1234567890124 | ethnicity | Mismatch | 31 | White British |
| Student | 1234567890125 | dob | Mismatch | 2004-02-19 | 19/02/2004 |
| Student | 1234567890126 | <record> | Missing in B | ∅ (absent) | ∅ (absent) |
| Enrolment | ENR-0003 | mode | Mismatch | 1 | FT |
| Student | 1234567890127 | <record> | Missing in A | ∅ (absent) | ∅ (absent) |
DHP overall=88.9% | Student=86.7%(3), Enrolment=91.7%(3)