ITR: Cache-Resident Databases

点击翻译按钮获取中文摘要

AbstractDatabases are at the very heart of the information economy. Databaseperformance is a driving factor that dictates what is possible through theuse of information technology. While database management systems haveevolved since they were invented several decades ago, their current designis unfortunately antiquated given the state-of-the-art computer memoryhierarchies of today (and even more so tomorrow). This project seeks toalleviate this problem.While processor speeds double every year, memory access speeds follow amuch shallower improvement curve. To bridge this speed gap, small, fastmemories called caches are used to hold frequently accessed data andinstructions close to the processor. When executing database workloads,accesses often miss in the (fast) cache and access the (slow) memory,thereby reducing performance significantly. Hardware approaches aretypically limited by access time constraints and by applicability to a widerange of workloads. To keep the hardware design feasible, caches typicallyuse simplistic data placement and replacement schemes, and are oblivious tothe memory access behavior of the application.Cache-conscious software methods are, on the contrary, extremely promising.The proposed algorithms collect data statistics in order to correctly groupdata with similar usage patterns and optimize cache utilization. Bycarefully observing behavior, data is prefetched into the cache before itis used. Preliminary results demonstrate that these techniques (i) minimizethe number of misses in the cache and (ii) significantly reduce theincurred penalties.