The Laboratory for Advanced System Software (LASS) (차세대 시스템 소프트웨어 연구실) is directed by Prof. Youngjae Kim and is part of Graduate School of Computer Science and Engineering at Sogang University. After being found in Fall 2016, LASS investigates research on computer systems, operating systems, distributed systems, parallel systems, file and storage systems, and data analytics systems. Specifically recent work by the LASS is pursued for research on computing systems with an emphasis on distributed file and storage systems, non-volatile memory technologies, and big data analytics platforms.
LASS (Laboratory for Advanced System Software) 연구실은 2016년 가을에 서강대학교 컴퓨터공학과 대학원에 설립되었으며 클라우드, 빅데이터, IoT 임베디드 시스템 플랫폼에 최적화된 시스템 소프트웨어 (운영체제, 미들웨어, 응용 소프트웨어) 연구를 수행합니다. 최근에는 차세대 컴퓨팅 시스템을 위한 분산 시스템 소프트웨어와 차세대 메모리 기반 스토리지 기술을 연구합니다.
- 미래창조과학부 ICT R&D 사업 초고성능 빅데이터 플랫폼 연구단 (연구수행기간: 2015-2019, 총 4년!) 참여 연구 기관 선정!
- 주관기관 울산과학기술원을 포함하여, KAIST, POSTECH, SNU 등 국내 우수 대학과 공동 연구를 수행합니다.
- 특히, LASS 연구실은 통합 분산 파일 시스템 및 지능형 데이터 관리 시스템 연구 개발을 수행하고 있습니다.
- SK Telecom 산학 연구 협약 - 연구주제: 클라우드 오브젝트 분산 스토리지, Ceph 데이터 중복제거 기술 연구
- 플래시 메모리 기반 클라우드 대용량 스토리지/파일 시스템 중복 데이터 제거 기술 연구를 수행합니다.
- SKT 국내 최고의 연구진과 공동 연구를 수행합니다.
- 연구재단 신진연구 과제 선정 (연구수행기간: 2015-, 3+2년!)
- 차세대 메모리/플래시 메모리 기반 빅데이터 스토리지 연구 수행하고 있습니다.
- Youngjae Kim
- Assistant Professor, in Computer Science and Engineering, Sogang University, Fall 2016-
- Research Interests: Distributed an Parallel Computing Systems, File Systems, Operating Systems, Resource management for Data Centers, Embedded Systems
- Awais Khan (PhD), 2015-
Antique Muhammad (PhD), in Computer Engineering, Ajou University (with Prof. Tae-Sun Chung) 2011-
- Area: Distributed File System
Donghyun Kang (MS), in Computer Science and Engineering, Sogang University
- Area: Cloud Computing, Performance Modeling
TaeUk Kim (MS), 2017-
- Area: Distributed File System, Solid-State Drive
Hyunki Byun (MS), 2017-Preethika Kasu (MS), in Computer Engineering, Ajou University 2016-
- Area: Distributed File System
Prince Hama (MS), in Computer Engineering, Ajou University 2016-
- Area: Fault tolerant data transfer protocol, I/O and Networking, Common Communication Interface
Ronnie Mativenga (MS), in Computer Engineering, Ajou University 2015-
- Area: Distributed File System, I/O and Networking
- Undergraduate Students / Interns
- Changgyu Lee (이창규), BS (senior) in Software and Computer Engineering, Ajou University 2013-
- Working for distributed file systems such as Ceph and Gluster
- Guncheol Lee (이건철), Department of Computer Science and Engineering, Sogang University
- Working with SSD Jasmine board
- Jinwoo Ahn (안진우), Department of Computer Science and Engineering, Sogang University
- Working with SSD Jasmine board
- Jeho Song (송제호), Department of Computer Science and Engineering, Sogang University
- Working with SSD simulator
- Heerak Lim (임희락), BS (senior) in Software and Computer Engineering, Ajou University, 2013-
- Embedded Systems, Solid-state Drive, Programming on Open SSD Jasmine Board
- Distributed File Systems
Performance of the file systems, have proved critical to the overall performance of an exceedingly broad class of applications. Traditional solutions, in which server exports a file system hierarchy that clients can map into their local name space, experience significant obstacle to scalable performance. Ceph, a distributed file system that provides excellent performance, reliability, and scalability. Ceph file system maximizes the separation between data and metadata management by replacing allocation tables with a pseudo-random data distribution function (CRUSH) designed for heterogeneous and dynamic clusters of unreliable object storage devices (OSDs). Our lab research aims at Data-deduplication and QoS schemes for Ceph file system. Also, develop Unified file system to aggregate different file systems to provide a single name space in virtual data facility in wide-area network environment, facilitating big data analysis. We also address data migration issue between tiers in Gluster FS, which is tiered storage systems using heterogeneous bricks (disks).
- Data Management
Data Management is one of the essential part for data-intensive scientific applications. In most of the cases, application performance is based on the efficiency of the data management algorithms. Also, it depends on the scalability of the data management system. Our lab research makes use of Key-value store and SciDB (Data Management System for Large Scale Scientific Data) as our main data management systems. Our aim is to optimize the queries in order to improve the performance of the applications.
- Data Transfer Application
One of the challenges with the big data is the data movement across the data centers geographically. Multiple bottlenecks exist along the end-to-end path from source to sink. Data storage infrastructure at both the source and sink and its interplay with the wide area network are increasingly the bottle neck to achieve high performance. Our lab research addresses these kind of challenges by developing a new bulk data movement framework called LADS (Layout-Aware Data Scheduling)for terabit networks. LADS exploits the underlying storage layout at each endpoint to maximize throughput without negatively impacting the performance of shared storage resources for other users. LADS uses the CCI (Common Communication Interface) in lieu of the sockets interface to use zero-copy, OS-bypass hardware when available. At present research is going on for extending LADS Frame Work to support fault tolerance.
- Image Storage Systems
Storing images in database is lot easier to manage than storing them in the file system. Of Course, the classic argument against the database is that it is slow. Retrieving images from database takes longer than retrieving an image from the file system. If the amount of the image data under consideration is small, then there is not much difference between the two methods. Considering the pace at which image data is growing in the recent past, need to address the problems associated with the file system in order to improve the Image Storage Systems.Our lab research focus will be mainly on building an image storage system for scalable training of deep neural networks.
- All Flash Storage Systems
NAND Flash memory based Solid State Drives (SSDs) exhibit many unique technical advantages over traditional hard disk drives (HDDs). Specifically, in terms of high random access performance and lower power consumption. On the other hand, the limited lifespan along with the associated reliability issues limit the use of SSDs in reliability-sensitive environments, such as data centers. In order to improve the high data access performance, Our research aims at I/O optimization for flash storage systems. Also, our research focuses on cache and Flash Translation Layer (FTL) to enhance the endurance of SSDs at the device level.
- I/O Optimization for Mobile Operating Systems
In most of the Mobile OS, file system and database are stacked vertically. Applications are developed using SQLite database queries. These queries are not optimized. For example, in Android, I/O operations to file systems are not optimized for performance. Our lab research aims at optimizing I/O operation for performance between such vertical I/O stack from application and storage device (NVRAM, Flash etc)
- Cluster System Administration
For distributed storage evaluation, we use a testbed which is composed of 9 servers including 1 management server attached with 28 hard drives. All these servers are interconnected via Gigabit Network Switch.
[CAL'17] Junghee Lee, Kalidas Ganesh, Hyuk-Jun Lee, Youngjae Kim*, FeSSD: A Fast Encrypted SSD Employing On-Chip Access-Control Memory, IEEE Computer Architecture Letters (IEEE CAL) (2017). (Accepted)
- [TPDS'17] Youngjae Kim, Scott Atchley, Geoffroy R. Valle ́e, Matt Lee, Galen M. Shipman, Optimizing End-to-End Big Data Transfers over Terabits Network Infrastructure, IEEE Transactions on Parallel and Distributed Systems (IEEE TPDS), Vol. 28, No. 1, pp. 188-201, January 2017. [IEEE Explore] (Impact Factor: 2.17, According to Thomson Reuters' 2013 Journal Citation Report)
- [PEVA'16] Seung-Hwan Lim and Youngjae Kim, A Quantitative Model of Application Slow-Down in Multi-Resource Shared Systems, Performance Evaluation (PEVA) (2016), http://dx.doi.org/10.1016/j.peva.2016.10.004 [pdf]
- [USENIX FAST'15] Youngjae Kim, Scott Atchley, Geoffroy R. Valle ́e, Galen M. Shipman,LADS: Optimizing Data Transfers using Layout-Aware Data Scheduling, In Proceedings of the USENIX Conference on File and Storage Technologies (USENIX FAST) (2015), San Jose, CA, February, 2015. (Acceptance Rate: 28/130 = 21.5%) [paper]
- [SC'15] Hyogi Sim, Youngjae Kim, Sudharshan Vazhkudai, Devesh Tiwari, Ali Anwar, Ali Butt, Lavanya Ramakrishnan, AnalyzeThis: An Analysis Workflow-Aware Storage System, In Proceedings of the 2015 ACM/IEEE International Conference on High Performance Computing, Networking, Storage and Analysis (SC) (2015), Austin, TX, November, 2015. (Acceptance Rate: 79/358 = 22.0%) [paper]