Skip to main content
GIFRÖST / Compare

Refine pair-level logic only where defaults are not enough.

The Advanced tab is where Compare stops being a broad setup flow and becomes a precise reconciliation tool. It gives detailed control over selected table pairs, data type alignment, SQL filtering, and delta-based comparison windows.

OverviewInspect all selected table pairs and their readiness.
Type handlingReview how source and target data types are normalized.
Pair-level tuningConfigure columns, SQL conditions, and comparison keys.
Delta modeCompare only the records that changed inside a time window.

What lives in Advanced

Section 01Tables Overview Table

Summarizes all table pairs currently selected for reconciliation and their validation state.

Section 02Data Types Table

Shows the type mappings used to normalize values before they are compared and hashed.

Section 03Compare Pair Dialog

Opens detailed controls for columns, SQL conditions, and delta-based processing on a single pair.

Tables Overview Table

This view keeps the full reconciliation scope visible in one place. It helps users remove, disable, validate, or inspect table pairs before the run starts.

FieldEnabled

Controls whether a given table pair is included in the reconciliation run. Invalid pairs can be disabled automatically.

FieldCompare Pair Name

Displays the schema-based pair identifier and opens deeper pair-level configuration.

FieldTable (Source)

Shows the source table together with estimated size and detected column count.

FieldTable (Target)

Shows the paired target table together with estimated size and detected column count.

FieldFetch Time

Reports how long metadata retrieval took, which helps expose slow objects or environments.

FieldActions

Lets the user remove a table pair entirely from the reconciliation setup.

ValidationPassed

Validation confirms compatible keys and column structures. The pair is ready for reconciliation.

ValidationWarning

Something requires review, such as partial column matching or another non-blocking inconsistency.

ValidationError

A critical issue such as missing keys or incompatible structures blocks the pair from running.

Data Types Table

Data type mapping exists to make the comparison fair. Values from different engines often need to be cast into the same canonical representation before they are hashed or compared value by value.

Data type mappings overview

The Data Types table defines how source and target types are aligned so the same business value produces the same comparison result.

Consistent castingEqual values stay equal after normalization
Source · Oracle NUMBER
42.50"42.50"a1b2c3
Target · PostgreSQL numeric
42.50"42.50"a1b2c3

Same canonical string and identical hash. Compare treats the row as consistent.

Inconsistent castingFalse discrepancies appear when rules diverge
Source · Oracle NUMBER
42.50"42.50"a1b2c3
Target · PostgreSQL numeric
42.50"42.5"9f8e7d

Different canonical output and different hash. The row looks inconsistent even though the business value can still be the same.

Common normalization pitfalls

Numeric precision Trailing zeros, scale, and rounding rules can alter canonical output.

Date and time values Formatting, time zones, and precision can cause false mismatches.

Character data CHAR vs VARCHAR, padding, and empty string handling need consistent rules.

Compare Pair Dialog

The dialog attached to a selected pair is divided into three tabs. Together they control column-level alignment, row filtering, and time-windowed comparison.

Tab 01Columns

Pair columns manually, define reconciliation keys, and exclude fields that should not participate.

Tab 02Conditional Processing

Filter rows through SQL conditions so only the relevant subset enters the comparison.

Tab 03Delta Processing

Compare only data that changed over time, which is ideal for cyclical or scheduled runs.

Columns

Key modeUse Table Primary Key

Automatically uses the primary key defined in the database.

Key modeUse Index

Lets the user pick an available index as the reconciliation key.

Key modeManual Columns Selection

Allows explicit manual selection of the key columns.

What users review in the Columns tab

Column Name (Source / Target) Shows the matched source and target fields, with PK labels where relevant.

Key flags Marks whether a column belongs to the reconciliation key according to the selected key mode.

Data type mapping Shows the original type and the mapped comparison type on each side.

Conditional Processing

SQL filteringRestrict rows before comparison

Users can choose a source or target column and define SQL conditions that decide which rows should enter the reconciliation scope.

ValidationTest the condition before running

Compare can validate the SQL expression and preview the rows that satisfy it, which reduces the chance of filtering the wrong slice of data.

InputSource and target row setsFull candidate population before filtering.
PredicateSQL conditionColumn-level WHERE logic restricts the scope.
OutputFiltered reconciliation setOnly the selected slice enters key resolution and comparison.

Conditional Processing acts before reconciliation proper. It narrows the data scope before hashing, key checks, and delta logic are applied.

Index hintUnique index

The selected column belongs to a unique index and is usually a strong candidate for filtering or key logic.

Index hintShared or multiple indexes

The column participates in several index paths and may need closer review.

Index hintNo index

The column is not indexed, which can affect performance and should be used deliberately.

Delta Processing

Delta Processing is built for repeated runs where scanning the entire table every time would be unnecessary. It focuses only on data changed within a selected time window.

Delta Processing flow

Delta mode uses a replicated timestamp column to limit the comparison to records changed between two selected points in time.

T0Start of the comparison range

Value source: User input, Get Value, or the previous T2

Run impact: Acts as the starting point for the current run.

T1End of the comparison range

Value source: User input or Get Value

Run impact: Defines the upper bound of the analyzed window.

T2Time when the comparison completed

Value source: Saved automatically by the system

Run impact: Becomes the next T0 in the following run.

BenefitAnalyze only what changed

Unchanged records stay outside the comparison window.

BenefitReduce unnecessary work

The run spends less time and infrastructure on stable data.

BenefitFit cyclical scheduling

The handoff from T2 to the next T0 makes recurring validation much easier to automate.