I’m trying to build a custom connector for Looker Studio due to existing connectors not being able to solve my use-case. My data is coming from BigQuery but aggregations need to happen on the SQL side (HLL.EXTRACT for example is not supported in Looker Studio calculated fields). Plus all my datasets need a large amount of calculated fields directly from my Semantic Layer, but even if Looker Studio calculated fields supported all necessary functions, I cannot programmatically create them without a custom connector.
Luckily, custom connectors exist. I got one working. However, it is extremely slow.
Even when all my external API calls come from AppScript cache (or when I hardcode them), and with a BigQuery SQL time of <1 sec, with a result size of like 2x2, it takes Looker Studio about 10+ seconds to render a graph.
Custom Connector Executions shows ~3 sec for a getData call, already longer than it should be, but not 10+ seconds. FWIW I’m using the BigQuery advanced service.
What can I do to make this fast? It seems like AppScript itself is the bottleneck but as far as I can tell there is no alternative. I’d love to host the connector myself or do whatever is necessary to make this as fast as it theoretically really should be able to be.