Help

The Community will be undergoing maintenance from Friday February 21 - Friday, February 28 and will be "read only" during this time. To learn more, check out our Announcements blog post.

Re: Factors Affecting Base Performance

72 0
cancel
Showing results for 
Search instead for 
Did you mean: 
Mike_AutomaticN
10 - Mercury
10 - Mercury

Hey there heavy-lifiting Airtable power users (i.e. full time consultants, developers and similar)!

I recently engaged in different discussions with fellow consultants on what factors affect base performance (e.g. speed) the most, trying to come up with a robust conclusion and hopefully a list of factors ordered by magnitude of impact (not the list below which is just me putting some of the factors out there to kick-off the discussion). 

•Number of Records: Probably super big impact?

•Number of Fields: Does having more fields significantly slow down the base? How much vs other factors?

•Number of Views: How much do multiple views -even if not opened- affect performance? vs other factos?

•Complex Formulas: Are complex or nested formulas a major culprit, especially if they reference many fields?

•Formulas on Rollups: Do rollup formulas or those within linked records introduce additional delays?

•Extensions: Do various extensions influence base responsiveness? Even if not used?

•Interfaces: Does the number of interfaces or pages within an interface affect overall speed?

•Automations: How do built-in Airtable automations impact performance compared to external automation tools (like Make or n8n) that interact via the API?

•Others. e.g. syncs.

I'd love to learn about different experiences with big bases.
Do you have very strict rules which you usually follow to avoid the base slowing down once it scales? Do you go about it strategically in advance?

Also, getting Airtable's team input would be amazing.

@ScottWorld  @TheTimeSavingCo @kuovonne @Alexey_Gusev @Pascal_Gallais- 

Mike, Consultant @ Automatic Nation

5 Replies 5
Kenneth_Raghuna
8 - Airtable Astronomer
8 - Airtable Astronomer

Following!

Alexey_Gusev
13 - Mars
13 - Mars

Hi,

I would merge the first two points into 'number of cells.' Sometimes I notice slow performance when the number of records exceeds 20K, but it also depends on the number of fields. In such cases, Airtable suggests adding some limitations, like hiding certain fields, and this significantly improves performance.

I can note that over time, Airtable has been handling large tables better, meaning performance is definitely improving gradually. Occasionally (less than once a month), there are performance issues with any tables, but it's usually a temporary, likely global, issue, that gets resolved fairly quickly. Usually it can be monitored via Airtable Status page

I tried adding a rather complex formula—a 'list of dates'—to a table with 72K records, calculating the final date from two other columns, and I didn’t see any issues at all when adding it.

Regarding views, sometimes a large table slows down if grouping is applied in the view.

As for extensions, if a default dashboard has many extensions and one of them affects performance, Airtable suggests pausing some of them. I usually check which extensions is causing the slowdown and move it to a different dashboard, which also solves the problem.

These are my impressions as a user.

Hi Mike,

My experience in terms of performances with Airtable is somehow erratic. Sometimes a very small database takes a very long time to open and sometimes a large one opens very quickly. It is not linked to specific databases, it varies from one day to another, so I guess the principal factor is the load on Airtable servers. 

I did observe recently that automations takes longer to launch. Automation duration remains the same, but the triggering phase seems slower than it used to be.

The other point that I have observered from the beggining is that when I need to modify an automation that implies an e-mailing action, reactivity when typing some new input in the actual mail to be sent is very poor.

Apart from this last point, I have never been faced with critical performances issues.

regards

Pascal

kuovonne
18 - Pluto
18 - Pluto

Based on anecdotal information, here are some of the things I believe affect base performance, in no particular order.

  • Use of NOW() in formula fields. This is my #1 thing to avoid. I also avoid TODAY() and other formula functions that recalculate based on the passage of time.
  • Formula fields and rollups with complex calculations, especially with rollups that pass data back-and-forth between tables multiple times with lots of linked records. 
  • Syncs. Both having a synced table and being a sync source.
  • Very high record counts.
  • How frequently a base is accessed. Infrequently used bases tend to be slower, especially on first load.
  • Heavy API usage changing data. Each API call asks Airtable servers to do something with the base.
  • Scripts that make a massive amount of data changes in a short period of time can temporarily slow down the base. Performance usually improves as soon as the script is done.
  • When an automation is triggered by thousands of records at once, it can take a long time to process all of them.
  • Having broken calculated fields (e.g. lookups where the source field has been deleted, formula fields that produce an error).
  • Open extensions that are resource hogs. (Only some are resource hogs.) An extension only runs if its dashboard is open. However, some extensions run continuously when they are open, watching and reacting to data changes. In general, I prefer workflows that a dashboard/extension only when it is needed.

Here are some things that I have not noticed affecting performance. 

  • Having lots of editable fields that are not used in filtering/sorting/grouping.
  • Having lots of data views with configurations that don't change
  • A complex formula that rarely needs to be recalculated (usually because none of the inputs change).
Mike_AutomaticN
10 - Mercury
10 - Mercury

Thanks @kuovonne @Pascal_Gallais- and @Alexey_Gusev for such interesting insights ðŸ”¥ðŸ”¥ðŸ”¥.
This is def something that I'd like to follow up on, hopefully sharing further insights and conclusions on this post.

Mike, Consultant @ Automatic Nation