I'm exploring whether it's a smart move to replace AWS QuickSight with Redshift and a React app for our data visualization needs. Currently, we're dealing with a setup that includes S3, Glue, Athena, and Postgres feeding into QuickSight, but we're running into significant latency issues. Our users often experience load times of 60-90 seconds, and queries can take anywhere from 8 to 30 seconds. This has become a notable pain point given our user base of 100-200 concurrent users.
After struggling with performance and high costs associated with QuickSight, I've experimented by mirroring our fact table into serverless Redshift, where the initial cold starts are eliminated, and query times have dropped to under 2 seconds—mainly due to API overhead. The existing approach has been less than ideal, and there's talk of transitioning to other tools, but I wanted to know if moving to Redshift backed by React is the right direction to minimize latency while also addressing user interface concerns without overspending.
3 Answers
I get your frustrations with the entire AWS data stack. I've seen similar issues, which is why I’ve switched to BigQuery on Google Cloud with Metabase as the front-end. Different platforms can definitely give you better options for performance and usability, so don’t shy away from exploring those alternatives!
Absolutely, moving to a proper data warehouse like Redshift can be a game changer. QuickSight has its flaws and can be cumbersome, especially for large datasets. You need something built for handling analytics, and from what you describe, QuickSight just isn't cutting it for you.
The current load times are concerning. If users start complaining about anything over 5 seconds, that’s a red flag. Think about whether the custom build will really provide enough speed improvements compared to the existing tool. Sometimes, it's about spotting the right latency threshold that makes sense for your users.

Totally, users have very low tolerance for slow responses. If they’re getting frustrated at just 5 seconds, it's vital to find a solution fast!