A concern with using LRU eviction is that the behavior is sensitive to how accesses are distributed over time. The useful property of random eviction versus LRU eviction is that performance falls within a narrow bound: best case scenario is worse, but worst case scenario is better. I think this use case is valuing predictability.
The main reason most databases use one of the myriad clock sweep eviction variants instead of true LRU eviction is that they ensure more uniform performance over all possible sequences of a given non-uniform distribution of accesses -- most of the upside of an LRU while (especially in clever variants) avoiding most of the downside as well.
The main reason most databases use one of the myriad clock sweep eviction variants instead of true LRU eviction is that they ensure more uniform performance over all possible sequences of a given non-uniform distribution of accesses -- most of the upside of an LRU while (especially in clever variants) avoiding most of the downside as well.