Keeping Aeion Fresh After the Initial Migration
A one-time migration answers "how do I get my data in." This page is about what comes after — when the source system (HubSpot, QuickBooks, a legacy ERP) keeps generating new records and you want Aeion to stay reasonably current without re-doing the whole import from scratch. Singularity doesn't run a background scheduler for you, but every job is one-directional (source → Aeion, read-only against the source), re-runnable, and safe to re-run repeatedly thanks to smart upserts — so "stay in sync" here means "re-run the same job on whatever cadence you want," not a live bidirectional pipeline.
What "Staying in Sync" Actually Means Here
`
Pattern — Re-Running an Import for a Phased Rollout
`
Pattern — Consolidating Multiple Sources Into One Aeion Record
`
Getting Aeion-Derived Data Back Out (What You'd Build Yourself)
`
FAQ
No. Singularity is an on-demand import engine — you (or your own external scheduler, calling the import API) decide when a job runs. There's no built-in CRON scheduler or watermark-based delta tracking inside Singularity itself.
Smart upserts. You pick a matching field when you configure the import (an external ID, an email address); each run checks the destination collection and updates the existing record on a match instead of inserting a new one.
No — every Singularity connector is one-directional (reads from the source, writes to Aeion). There's no bidirectional sync and no reverse-ETL feature. Getting data out of Aeion into another system is a custom integration against Aeion's own API.
There's no automatic schema-drift detection. If a source adds, renames, or removes a field, you'll need to revisit the field mapping yourself — the dry-run preview (processing the first 500 rows) is the fastest way to catch a mismatch before a full re-run.
Yes, via the generic REST/GraphQL connector — included at no extra cost — with OAuth or API-key auth and custom pagination — you configure it once and it behaves like any other connector on re-run. For SOAP or truly proprietary protocols, that requires a custom connector.
As fast as you're willing to trigger the job. If your own scheduler calls the import API every 5 minutes, that's your effective sync interval — the constraint is your own automation and the source system's rate limits, not anything Singularity throttles down to a slower built-in cadence.
All connector credentials are encrypted at rest per-tenant. All traffic is TLS. Field-level PII redaction is available during mapping.
Since there's no built-in scheduler, "pausing" just means not triggering the job — whether that's clicking nothing in the admin UI or turning off your own external cron.