A5066717095- Advanced Data Storage Technologies
- Caching and Content Delivery
- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Cloud Computing and Resource Management
- Algorithms and Data Compression
- Peer-to-Peer Network Technologies
- Distributed systems and fault tolerance
- Cellular Automata and Applications
SK Group (South Korea)
2023-2024
SK Group (Japan)
2019-2023
SK Group (United States)
2022-2023
Research Triangle Park Foundation
2002
Several key-value stores such as RocksDB and MongoDB are implemented on the file system using Log-Structured Merge-Tree (LSM-tree). The LSM-tree involves high compaction overhead. To minimize this overhead, WiscKey, state-of-the-art LSM-tree, separates key value, appends value to Value Log file, manages only offset. This minimizes overhead by reducing number of SSTables managed LSM-tree. However, WiscKey still has a I/O stack that must go through OS block-layer. Therefore, paper proposes...
Popular software key-value stores such as LevelDB and RocksDB are often tailored for efficient writing. Yet, they tend to also perform well on read operations. This is because while data initially stored in a format that favors writes, it later transformed by the DB background into better accommodates reads. Write-optimized can still block writes. happens when those workers cannot keep up with foreground insertion workload.This paper advocates hardware-accelerated store, enabling...
Key-Value SSD (KVSSD) has shown great potential for several important classes of emerging data stores due to its high throughput and low latency. When designing a key-value store with range queries, an LSM-tree is considered better choice than hash table key ordering. However, the design space queries in LSM-tree-based KVSSDs yet be explored, despite being one most demanding features. In this paper, we investigate constraints from perspective propose three principles. Based on these...
In this paper, we propose OctoKV, an innovative network-based key-value storage system. OctoKV addresses the repetitive address translation overhead associated with traditional stores running on file systems client side. To mitigate overhead, implemented store server side using NVMe-oF and a user-level NVMe driver. particular, employed fine-grained resource monitoring load balancing based heuristics to optimize I/O performance. is deployed Linux cluster Intel SPDK. The extensive evaluation...
The Key-Value Solid State Drive (KV-SSD) represents a significant evolution in storage device interfaces by accommodating non-page-aligned key-value pairs, departure from conventional models. However, KV-SSDs encounter challenges as their specialized data transfer and packing requirements conflict with established protocols like NVMe, which are designed around fixed memory page units. This discord leads to inefficient movement increased NAND write I/Os, turn escalates network traffic...
When data is loaded from a key-value store to the GPU in conventional GPU-driven computing model, it entails overhead of all heavy I/O stacks and file system. This paper presents GPUKV, framework that eliminates aforementioned with less host-side usage resources such as CPU memory. GPUKV has following three features: (i) provides abstraction GPU; (ii) In when loading GPU, performed through PCIe peer-to-peer (P2P) communication without copying user kernel space memory; (iii) uses KVSSD, which...
Porting a distributed application from one environment to another can be significant, particularly when performance is an important consideration. The Andrew File System (AFS) file system designed heterogeneous and scalable, it runs efficiently on variations of Unix. A port AFS Operating system/2 (OS/2) encountered assortment problems at various levels. port, performed as sequence two ports, first investigated feasibility issues then integrated into the OS/2 environment. Additionally,...
Key-value SSDs (KVSSDs) introduced so far are limited in their use as an alternative to the key-value store running on host due following technical limitations. First, they were designed only for a single tenant, limiting of multiple tenants. Second, mainly focused designing indexes primary key-based searches, without supporting various queries using combination key and non-primary attribute-based searches. This article proposes Cerberus , Log Structured Merged (LSM) tree-based KVSSD armed...
A computational storage device (CSD) using Intel SPDK guarantees low latency and high throughput. The CSD must aid background tasks for the service applications (background tasks) without harming user I/O performance (foreground I/O) since is also a device. However, in practice, often increases foreground latencies underutilizes CPU cores CSD. These problems proceed from allocating I/Os to same core because processes them as request distinguishing them. To tackle this, we propose Background...