News for hackers. Stuff that matters.
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by zeristor on Thursday November 13, 2025 @04:17PM (#45920672)
We modelled using Redshift scheduling automation, that suggested a welcome 20% improvement.

We applied it and it run 20% slower.

As it runs it is supposed to learn how to optimise scheduling. Not sure how it gets over the burst at month, quarter, and year end.

Re:Turns Out AI Is Not Good at Database Transaction Scheduling by accheng on Thursday November 13, 2025 @04:38PM (#45920947)
What did you use to try to optimize scheduling?
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by qat321 on Thursday November 13, 2025 @04:41PM (#45920980)
turns out DB transactions are a hard problem...
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by taqpos on Thursday November 13, 2025 @04:42PM (#45920993)
I wonder what the optimal for these workloads are?
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by pos456 on Thursday November 13, 2025 @04:44PM (#45921009)
For the offline problem, AI found a new solution that was 34% faster. Interesting results!
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by gfhsad on Thursday November 13, 2025 @04:45PM (#45921021)
What about other use cases
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by accheng on Thursday November 13, 2025 @04:50PM (#45921087)
Check out our blog series here: ucbskyadrs.github.io!
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by kap901 on Thursday November 13, 2025 @04:46PM (#45921034)
Seems like AI hype
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by pakt1 on Thursday November 13, 2025 @04:48PM (#45921047)
Transactions seem like a hard problem for scheduling, wonder how the authors got this to work in the original paper?
Re:Turns Out AI Is Not Good at Database Transaction Scheduling by accheng on Thursday November 13, 2025 @04:50PM (#45921073)
As described in our VLDB paper (https://www.vldb.org/pvldb/vol17/p2694-cheng.pdf), we had to implement a number of optimizations and integrate scheduling with CC to get good performance.