Sharp analysis on the operational vs semantic gap in REST specs. The ListTables sync pathology you describe is exactly what breaks at scaleexcept most teams blame the catalog implementation rather than the underspecified protocol. I've debugged similar issues where query planners stall because there's no bounded expectation for metadata latency, so they can't distinguish between a slow catalog and a broken one. The commit contention problem is worse than people realize becuase aggressive retry clients don't just starve others, they amplify load during exactly the moments when the system is already stressed.
Sharp analysis on the operational vs semantic gap in REST specs. The ListTables sync pathology you describe is exactly what breaks at scaleexcept most teams blame the catalog implementation rather than the underspecified protocol. I've debugged similar issues where query planners stall because there's no bounded expectation for metadata latency, so they can't distinguish between a slow catalog and a broken one. The commit contention problem is worse than people realize becuase aggressive retry clients don't just starve others, they amplify load during exactly the moments when the system is already stressed.
True that. I hope the community addresses these issues.
This was interesting
Excellent points.
Interesting observation thanks
Thank you! Very helpful!