The use case used for this Friction Log covers an actual requirement to build analytics dashboards and insight across a 40+GB MongoDB dataset. Rick was able to do his prototype, end-to-end in about a day, on top of Rockset. The user, Rick has some data and analytics experience but isn't a deep data engineering junkie.
The SQL query language is easy to use, google if you run into problems, etc. There really was minimal learning overhead my data was in Rockset.
Importing a dataset from Mongo was very straightforward, from a first-time / try it out experience.
The time to ingest data, write a query, integrate it into an API, and then a React app was very fast. I was able to try things out on the fly, change on the fly, etc.
Rockset analyzed my queries and provided helpful ways to optimize it, which most of the time really improved the query time.
There are a lot of caveats to know if you want to really nail down a good ingest. For example, mapping a timestamp field is really important. The interface didn't always tell me that.
At one point the ingest froze at 55% and there wasn't anything I could do to fix it. Support was timely and helpful and ultimately the solution was to re-ingest from scratch.
At first, I presumed lambda queries allowed the system to further optimize the results but in the end, it just seems to be a way to memorialize a query in an endpoint.
Many times the editor's linter would be wrong or show an error when there wasn't one.
Sign up and get the latest friction log studies, blog articles and podcast updates