Share your ideas

Asynchronous Search Results - Return Results while Search Running

The current Tealeaf search service is synchronous, meaning it scans all canisters before returning results to the user. This is exceptionally time consuming, as well as can often result in timeouts to the user if they are broad results, or longer scans for data.

For larger Tealeaf systems, the search service, this happens frequently, having to scan many canisters for results and wait for them all to return. This can mean anywhere from 20-100+ canisters for very large systems. Running synchronously, this is VERY bad experience for users.

Newer data systems such as Splunk, Amplitude, Hadoop, etc are made to run async results, that as the search service finds results, it returns them in batches back to the Portal and displays them to the user in segments, constantly adding to the list of results.

This allows users to start working with the results as they wait for the search to complete, if there are already results done. If there are NO results returned, the batches back to the user still report some sort of progress: like how many canisters are scanned, what percentage of scanning - some kind of progression report so the user knows the system is moving, and to what extent. Silence in this new world of SPAs means users get impatient and blame a "slow system" or "poor performance" if they don't actually perceive the system as working hard, or over a large data set. Communication is key.

Thus leading to the request: To make Tealeaf On Premise search service and Portal display for search results ASYNC - to move all search results from a simple pop up to the search page that starts returning Replay session results for users to work, or communicate that the search is executing - including any timeouts or canister errors, reported inline/at time reported. This should also still include an option to STOP the search if they already found what they need.


  • Dieter Davis
  • Feb 10 2022
How will this idea be used?

This will increase the efficiency of our users to spend less time searching and more time replaying. Additionally, it will passively advertise to users for larger Tealeaf systems what level of sophistication exists under the hood. How many canisters it's searching, how much data it's scanning. Tealeaf can be a very LARGE system, and we can use this as an opportunity to advertise to a user how well it does, what it's doing, what kind of data swath it's scanning. Thus I recommend the search results be a little more verbose in it's status to impress this on the users.

What is your industry? Financial Markets
What is the idea priority? High