Improving storage I/O performance with active SSDs능동형 솔리드 스테이트 드라이브를 이용한 저장장치 입출력 성능 개선

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 554
  • Download : 0
In the last decades, there has been an explosion in the volume of data to be processed by data-intensive computing applications. As a result, processing I/O operations efficiently has become an important challenge. Although SSDs (solid state drives) can be an effective solution because of the fast I/O performance, there are still the overheads of I/O in the application and SSDs. Using high-level semantics can reduce the overhead of I/O in the SSD, however the simple block-level interface which is widely used for storage devices hinders the optimization of the SSD. The concept of in-storage processing can remove unnecessary data transfer and reduce the I/O overhead of the application by offloading a part of the data-processing tasks. This dissertation introduces intelligent flash-based storage device, called active SSDs, which utilize high-level semantics and in-storage computing to improve the I/O performance for data-intensive computing. Active SSDs improve the performance and reliability of SSDs using the high-level information transferred via the object-based interface which provides general abstraction layer to inform the properties of data. To support the object-based interface, this paper redesigns and implements the full software stacks on an ARM-based storage prototype. The FTL optimization with several object-aware schemes reduces the overhead of garbage collection inside SSDs. Active SSDs also improve the performance of the Hadoop application using a proposed in-storage processing scheme called ActiveSort. ActiveSort is a novel mechanism using the concept of in-storage processing to improve the external sorting algorithm which is used extensively in the data-intensive computing frameworks such as Hadoop. By performing merge operations on-the-fly within the SSD, ActiveSort reduces the amount of I/O transfer and improves the performance of external sorting in Hadoop. Several evaluations on the real prototype indicate that active SSDs improve the performance, energy efficiency, and the lifetime of SSDs for various applications.
Advisors
Maeng, Seungryoulresearcher맹승렬researcher
Description
한국과학기술원 :전산학부,
Publisher
한국과학기술원
Issue Date
2016
Identifier
325007
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 전산학부, 2016.8 ,[v, 68 p. :]

Keywords

NAND flash memory; solid-state drives; object-based interface; in-storage processing; MapReduce; 낸드 플래시 메모리; 솔리드 스테이트 드라이브; 객체 기반 인터페이스; 저장장치 내 처리; 맵리듀스

URI
http://hdl.handle.net/10203/222395
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=663214&flag=dissertation
Appears in Collection
CS-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0