Skip to main content
GIFRÖST / Compare

Set the default behavior before large runs hit the infrastructure.

Global Settings are available throughout the reconciliation setup and control how Compare behaves across the whole workflow. They are split into Common, Agent, and Server areas so tuning can happen at the right layer.

Always availableThe tab can be reached during every setup step.
CommonControls comparison behavior, retries, and safety limits.
AgentCovers execution close to local or remote agents.
ServerCovers settings at the central coordination layer.

Settings areas

Area 01Common

Defines the general comparison parameters that affect reconciliation behavior end to end.

Area 02Agent

Contains parameters specific to the local or remote agent participating in the run.

Area 03Server

Contains settings applied at the server layer that coordinates and manages the process.

Entry pointGlobal Settings tabOne control surface available throughout the workflow.
CommonComparison defaultsBatching, retries, error limits, and normalization rules.
Agent / ServerExecution and orchestration layersLocal runtime behavior and central coordination share the same configuration entry point.

Conceptually, Global Settings behaves like a control plane: one tab fans out into comparison defaults, agent execution behavior, and server orchestration rules.

Common settings

The Common section contains the most operationally important defaults because it controls throughput, retry behavior, connection pressure, and failure thresholds.

SettingBatch Size

Number of records processed at once during table comparison.

SettingRecheck keys wait durations

Determines how long Compare waits before rechecking rows that initially looked different.

SettingMax database connections

Maximum number of concurrent database connections. The value must be greater than 0.

SettingMax errors

Maximum accepted number of differences before Compare stops a table run. The value must be greater than 0.

SettingRecheck keys

Enables a second verification pass for rows marked as different.

SettingTrim Trailing Spaces

Removes trailing spaces from text values before comparison.

Runtime axisThroughput envelope

Batch Size and Max database connections determine how much pressure the run puts on the environment.

Runtime axisConsistency tolerance

Recheck keys and the wait duration shape how aggressively Compare treats transient replication drift.

Runtime axisFailure guardrails

Max errors and normalization helpers such as Trim Trailing Spaces reduce noise and stop obviously broken runs earlier.

What to tune first

PriorityThroughput

Start with Batch Size and Max database connections when the main question is how much load the environment can sustain.

PriorityFalse-positive control

Use Recheck keys and the related wait duration when live replication may temporarily make rows look out of sync.

PrioritySafety limits

Set Max errors deliberately so the job stops when noise turns into a real issue instead of burning time on a broken run.

Agent and server focus

Execution layerAgent settings

Agent-specific parameters matter most when comparison logic runs close to the database and local execution characteristics influence performance.

Coordination layerServer settings

Server-level settings matter when orchestration, central resource handling, and global service behavior need to be aligned with the run profile.

Practical recommendation

Start conservative Use safe connection and error limits before increasing throughput.

Tune to workload Cyclical delta runs and full one-off audits usually need different defaults.

Review after first production runs Real fetch times and discrepancy patterns are the best source of tuning decisions.