From the docs, the Instagram source does not automatically set the start date to “today” when you create the connector. - The Instagram connector has an optional Start Date field. If you fill it, “all data generated after this date will be replicated.” [[Instagram source](https://docs.airbyte.com/integrations/sources/instagram)] - If you leave it blank, Airbyte automatically sets the start date to 2 years before the present date, not to today. [[Instagram source](https://docs.airbyte.com/integrations/sources/instagram)] So if you create the connector on the 19th and don’t set a start date, it will start from “today minus 2 years,” not from the 19th itself. If you want it to start exactly from the 19th, you must explicitly enter that date in the Start Date field.

Given what’s documented for the Instagram source, setting the Start Date to the connector creation date has a few important implications: 1. What the Start Date actually controls - The Start Date is optional; if you set it, “all data generated after this date will be replicated.” [[Instagram setup](https://docs.airbyte.com/integrations/sources/instagram)] - If you leave it blank, Airbyte defaults it to 2 years before “now”. [[Instagram setup](https://docs.airbyte.com/integrations/sources/instagram)] 2. Why your sync can still be long even with a “today” start date The knowledge sources indicate some current limitations in the Instagram connector: - The

start_date

is effectively used only for **User Insights**; media and stories are still fetched in full. [[Instagram improvements](https://github.com/airbytehq/airbyte/issues/61481)] - So even if you set the Start Date to the creation date, the connector may still: - Download all media and stories for the account(s), and then - Download insights for them (with incremental behavior only on some streams). [[Instagram improvements](https://github.com/airbytehq/airbyte/issues/61481)] - On top of that, Instagram’s API has strict rate limiting, which slows down large or repeated calls. [[Instagram limits](https://docs.airbyte.com/integrations/sources/instagram#limitations--troubleshooting)] 3. What this means for your case So even though you set the Start Date equal to the creation date, the long runtime is still expected because: - Start Date does not currently limit all streams (especially media/stories). [[Instagram improvements](https://github.com/airbytehq/airbyte/issues/61481)] - Rate limiting from Instagram/Facebook can make those full pulls slow. [[Instagram limits](https://docs.airbyte.com/integrations/sources/instagram#limitations--troubleshooting)] If you can share which Instagram streams you enabled (e.g., media, stories, user_insights), I can map more precisely how your Start Date is (and isn’t) being applied according to these docs.

Based on the available information, there’s no explicit guidance saying you should prefer EC2/self‑managed over Airbyte Cloud for Instagram, and there are no hard performance benchmarks comparing Cloud vs self‑managed for this specific connector. The knowledge sources mainly show: - Instagram syncs can be slow due to API rate limits and how the connector pulls data (all media/stories + insights), regardless of where Airbyte runs. [[Instagram limits](https://docs.airbyte.com/integrations/sources/instagram#limitations--troubleshooting); [Instagram improvements](https://github.com/airbytehq/airbyte/issues/61481)] - Users have seen environment‑dependent performance differences (e.g., local vs GCP), but that’s for other connectors (Google Analytics, MySQL), not Instagram specifically. [[Improving speed](https://discuss.airbyte.io/t/2752); [MySQL performance](https://discuss.airbyte.io/t/1092)] So, from the docs/issues: - Running on EC2 might behave differently (network, CPU type, etc.), but there’s no clear evidence it will reliably be faster for Instagram. - The main bottlenecks (Instagram rate limiting and connector design around

start_date

) will affect you in both Cloud and self‑managed. [[Instagram limits](https://docs.airbyte.com/integrations/sources/instagram#limitations--troubleshooting); [Instagram improvements](https://github.com/airbytehq/airbyte/issues/61481)] If you already have an EC2 environment, you could run a test sync there and compare total runtime vs Cloud, but the knowledge sources do not guarantee that EC2 will solve the long‑running syncs.

The knowledge sources don’t contain a direct, apples‑to‑apples comparison between: - Writing your own Python sync script for Instagram, vs. - Using Airbyte (Cloud or self‑managed) So I can’t definitively say which will be faster in your exact case. I can only outline trade‑offs that are implied by the docs and forum posts. ### 1. Performance: will a custom Python script be faster? The main bottleneck for Instagram is the Instagram/Facebook API itself (rate limits and how much data you ask for), not just Airbyte overhead. [[Instagram limits](https://docs.airbyte.com/integrations/sources/instagram#limitations--troubleshooting)] A hand‑written Python script might: - Avoid some framework overhead. - Be tuned very specifically to just the endpoints/fields you need. But you cannot bypass: - Instagram’s rate limits. - The cost of fetching large volumes of media/insights. The docs and forum posts do show that some users perceive Airbyte connectors as slower than “raw” HTTP calls in custom code, but there is no concrete benchmark or guarantee that a custom script will be “a lot faster” for Instagram specifically. [[Performance tuning custom connector](https://discuss.airbyte.io/t/4215)] ### 2. What benefits you lose by not using Airbyte Airbyte gives you, out of the box: - Incremental sync framework (state handling, at‑least‑once delivery, etc.). [[Python CDK incremental](https://discuss.airbyte.io/t/5259)] - Schema handling, testing, and deployment via the CDK and/or Connector Builder. [[Custom Python connector tutorial](https://docs.airbyte.com/platform/connector-development/tutorials/custom-python-connector/concurrency); [Custom components](https://docs.airbyte.com/platform/connector-development/connector-builder-ui/custom-components)] - Retries, error traces, logging, and UI to manage connections and jobs. [[Custom Python connector tutorial](https://docs.airbyte.com/platform/connector-development/tutorials/custom-python-connector/concurrency)] If you write your own Python sync from scratch, you must implement and maintain all of this yourself: scheduling, retries, backfills, schema evolution, state, monitoring, etc. The docs emphasize that the CDK and Connector Builder exist precisely to avoid re‑implementing these patterns. [[Custom components](https://docs.airbyte.com/platform/connector-development/connector-builder-ui/custom-components)] ### 3. Middle ground: custom connector instead of standalone script Instead of a completely separate Python script, the docs suggest: - Building a custom Python connector with the CDK when existing connectors don’t meet your needs. [[Webflow connector tutorial](https://airbyte.com/tutorials/extract-data-from-the-webflow-api#step-5)] - Or using Connector Builder + Custom Components to override only the parts you need (e.g., special pagination, transformations) while keeping the Airbyte framework. [[Custom components](https://docs.airbyte.com/platform/connector-development/connector-builder-ui/custom-components)] This gives you more control (and possibly some performance tuning) while still benefiting from Airbyte’s orchestration and reliability. --- Summary - A pure Python script might be somewhat leaner, but the sources do not show clear evidence that it will be “much faster” or “fetch a lot more data easily” than Airbyte for Instagram; you’re still bound by Instagram’s API limits. [[Instagram limits](https://docs.airbyte.com/integrations/sources/instagram#limitations--troubleshooting)] - You would give up Airbyte’s built‑in state management, retries, schema handling, and UI, and you’d have to rebuild those pieces yourself. - If Airbyte’s Instagram behavior is the main issue, the documented path is to customize via the Python CDK or Connector Builder (with Custom Components), not necessarily to abandon Airbyte entirely. [[Webflow connector tutorial](https://airbyte.com/tutorials/extract-data-from-the-webflow-api#step-5); [Custom components](https://docs.airbyte.com/platform/connector-development/connector-builder-ui/custom-components)]