Bϵ-樹在寫入密集的情況下，是一個很好的選擇，他將多個寫入請求合併到緩衝區， 這有助於減少寫入放大。SplinterDB [1] 是以Bϵ-樹實作而成，且配合了NVMe 裝置，然 而，他並不是為了開放通道式固態硬碟，我們提出了高度平行Bϵ-樹，他可以很好的利用 開放式固態硬碟有，而傳統固態硬碟沒有的優勢，在我們的論文中，提出一個方法，藉由 維持高度平行讀取，使得讀取延遲下降。;Bϵ-tree is a good candidate for write-intensive situations. It gathers multiple write requests into its buffer. This helps reduce write amplification (WA). SplinterDB [1] is implemented using Bϵ-trees and is compatible with NVMe solid-state drives[1]. However, it’s not designed for open-channel SSDs[2]. We designed HP-Bϵ-tree to leverage the advantages of open-channel SSDs over traditional SSDs. In this paper, we propose an approach to reduce read latency by maintaining high degree of parallel reads.
