Using HyperDB in Airtable

Plan availability	Enterprise Scale
Permissions	Owners/Creators - To create or update synced tables, you'll need to have creator permissions in the base where you are setting up the sync.
Platform(s)	Web/Browser, Mac app, and Windows app
Related reading	Getting started with Airtable sync - If you haven’t set up a sync before, we recommend first reading this article to become familiarized with how the Airtable Sync feature generally works. Airtable Sync FAQs - Answers to common questions that might help unblock you if you are encountering issues.

Airtable allows you to use data stored in HyperDB within a base in a few different ways. Builders can add HyperDB data to a table via on-demand sync or filtering data for specific information that's relevant to your needs.

	On-demand sync	Filtering
Best of	Performance - only add selected records without bogging down your base performance	Completeness - having all of the data in the table
Linked records	Linking data set to other tables in the base is important	I don’t need linked records, I just need to see all the records
Updates	Data will get dynamically updated from the HyperDB data set	Data will get dynamically updated from the HyperDB data set
Upon creation	A new empty table will be created. As records are linked to the HyperDB data set, they will populate in the new table	A new table will be created with all of the filtered records
Example	I want to bring in “Ratatouille”, “Cars”, and “Up” from my Movies data set	Let me bring in all Pixar movies from 2000-2010 in my Movies data set

What does it mean to pull in records on-demand?

On-demand sync allows you to link records as needed to a massive data set powered by HyperDB—giving you the ability to search millions of records without impacting performance.

On-demand syncs each record to your table as it is linked instead of the typical sync method, which brings in all records from a synced data set.

To get started, you need an existing table that you want to link to this massive data set. For example, in the diagram below, Zelos Inc. has an existing "Support tickets" table and wants to link the field "Requestor” to the HyperDB-backed data set "All users."
The user adds the "All users" data set as a new table, "Requestor," by choosing this data set from the data library. During setup, they bring data with on-demand sync, linking it to the table and field "Support tickets" and "Requestor," respectively.

Once the existing table is connected to the synced on-demand table, you can search for any record with the linked record picker. Only linked records will sync into the new table. In the example, support tickets can be assigned to any user in the "All users" dataset. The on-demand synced "Requestor" table only displays the subset of users that have been actively linked to tickets.

Note
You can remove unreferenced records from the table by choosing to delete unlinked records in the table settings.

What fields can I search using the linked record picker?

HyperDB requires that each record have a unique primary key. Because that primary key is often not easy to read, we’ve introduced the concept of a “search field” for HyperDB. The admin configures this and publishes it as part of the data set.

If the search field is visible in the linked record picker (the field cannot be hidden), the linked record picker will allow you to search by either the primary field or the search field.

As records are added with the linked record picker, they are added to a new, on-demand synced table:

How can I use automations to assist with linking?

Automations can be used to link to any record in a synced on-demand table as long as you know the unique ID for the record. You can do so by triggering an "Update Record" action in the local table that links to the on-demand synced dataset. Update a specific linked record field with the unique ID from the larger dataset, and it will get synced into the base automatically.

FAQs

Can I edit records in a synced on-demand table?

No, records can not be edited from how they appear in the original data set. However, you can remove unreferenced records from the table.

Can I search any column in the linked record picker using on-demand sync?

No, you can search through two columns—the primary key and an additional search key—to find records.

Can I preview the larger data set an on-demand sync pulls from?

No, you currently can't preview the data set besides using the linked record picker to scroll through records.