# High Throughput Push Based Storage Manager

## TL;DR

This paper introduces a push-based storage manager with prefetching and caching strategies that optimize I/O throughput and reduce read times in database systems, especially effective with SSDs.

## Contribution

It proposes a novel push-based prefetching strategy with cache management and query request merging to enhance storage I/O efficiency.

## Key findings

- Reduced total read time on SSDs
- Minimized I/O requests through caching
- Improved throughput with query request merging

## Abstract

The storage manager, as a key component of the database system, is responsible for organizing, reading, and delivering data to the execution engine for processing. According to the data serving mechanism, existing storage managers are either pull-based, incurring high latency, or push-based, leading to a high number of I/O requests when the CPU is busy. To improve these shortcomings, this thesis proposes a push-based prefetching strategy in a column-wise storage manager. The proposed strategy implements an efficient cache layer to store shared data among queries to reduce the number of I/O requests. The capacity of the cache is maintained by a time access-aware eviction mechanism. Our strategy enables the storage manager to coordinate multiple queries by merging their requests and dynamically generate an optimal read order that maximizes the overall I/O throughput. We evaluated our storage manager both over a disk-based redundant array of independent disks (RAID) and an NVM Express (NVMe) solid-state drive (SSD). With the high read performance of the SSD, we successfully minimized the total read time and number of I/O accesses.