I am a Ph.D. candidate at Renmin University
of China , Beijing. I got my Bachelor and Master's degree from Beijing Institute of Technology in 2002 and
2005. After graduation, I had worked in ChinaSoft International CO.,LTD as a vendor of Microsoft Corporation for two years. Since then, I have been doing research in Database field especially in
Flash Memory Database, under the Advisor of Prof. Xiaofeng Meng.
Education
- Ph.D. Candidate, September 2006 - present
Dept. of Computer Science, School of Information, Renmin University, Beijing, P.R. China
- M.S., September 2002 - January
2005
Dept. of Optical Engineering, School of Information, Beijing Institute of Technology, Beijing, P.R. China
- B.S., September 1998 - June
2002
Dept. of Optical Engineering, School of Information, Beijing Institute of Technology, Beijing, P.R. China
Research
Research Interests: Flash memory database;
Hardware-aware Storage, Indexing, Query processing and Transaction processing; Design, implementation, and performance of data management systems.
Research topic: Flash Memory Database. Compared with Hard Disk Driver (HDD), flash memory has a lot of advantages, such as high random read speed. With the increase of capacity and decrease of the price,
Solid State Driver (SSD) is envisioned to replace HDD as the secondary storage media.
- Storage
Besides read and write operations, flash memory has another data operation which is called ERASE. As for flash memory, erase has three main characteristics.
1. The granularity of read and write is page, while that of erase is block which usually contains 32 or 64 pages. 2. Erase has high cost. The speed ratio of read and write to erase is about 60X and 7.5X according to SAMSUNG datasheet.
3. Flash memory has limited erase times. Therefore storage management proposal should overcome the limitation of erase while taking full advantage of high random read speed. One way is to laten and merge write. The other way is to reduce garbage.
- Indexing
Out-of-place update is adopted by flash memory because of erase. The challenge brought by out-of-place update is cascade update which leads to low update performance and large quantity of garbage. According to this design principle, flash-based indexing should decrease write operations and merge small write. Besides this, flash-based indexing also should take full advantage of high random read speed of flash memory.
- Query Processing
In disk-based DBMS, tuples are arrayed
sequentially in most case. In this way query processing can scan tuples sequentially, so the DBMS gets high efficiency. However, as for flash memory, the speed of random read is similar to that of sequential read. Therefore, the limitation of scanning tuples sequentially does not exist in flash-based query processing. On the other hand, disk-based query processing writes large quantity of temp data in order to reduce the quantity of data to read. However, as for flash memory, write is a high cost operation and should be avoided as far as possibly. In a word, flash-based query processing should avoid the write operation as far as possibly and take full advantage of random read operation.
- Transaction Processing
Out-of-place update leads to multiple versions of data. This characteristic cater to
the MVCC.
Besides this, MVCC can read different versions of data with the help of the high random read performance. During the course of recovery, multiple versions of data also is very ussful. Some redo and undo
maybe can be avoided, therefore the efficiency of recovery can be enhanced.
Experiences
-
2009.01 – present Leading member of Flash-based Database Research (Key Project) granted by the Natural Science Foundation of China (NSFC) under grant number 60833005. This project aims to redesign the conventional disk-based database according to the characteristics of flash memory. In this project, I focused on flash-aware storage and index at first, then query processing and transaction processing.
-
2007.01 - 2008.01 Leading member of Flash-based Server Systems granted by Huawei Science Foundation, 2007. This project develops high-performance flash-based servers. In this project, I, as the member from Renmin University of China, designed the storage method and indexing structure.
-
2006.10 – 2006.12 Technical Manager of Syntech project granted by CVIC Software Engineering Co., Ltd, 2006. In this project, I design the database for the direct distributor relation and work flow.
Conference
- Da Zhou, Xiaofeng Meng: RS-Wrapper: Random Write Optimization for Solid State Drive. In Proceedings of the 18th ACM Conference on Information and Knowledge Management(CIKM2009), November 2-6, 2009, Hong Kong, China. (EI Index) [
pdf
| ppt ]
- Da Zhou, Xiaofeng Meng, Zhichao Liang: HF-Tree: An Update-Efficient Index for Flash Memory. In Proceedings of the 26th National Database Conference(NDBC2009), October 15-18, 2009, Nanchang, China.(In Chinese)
(EI Index)
[ pdf
| ppt ]
- Zhichao Liang, Da Zhou, Xiaofeng Meng: Sub-Join: A Query Optimization Algorithm for Flash-based Database. In Proceedings of the 26th National Database Conference(NDBC2009), October 15-18, 2009, Nanchang, China.(In Chinese)
[ pdf
| ppt ]
- Da Zhou, Zhichao Liang, Xiaofeng Meng: A New Cache Management Approach for Transaction Processing on Flash-based Database. In Proceedings of the 6th Web Information Systems and Applications Conference(WISA2009), September 18-20, 2009, Xuzhou, China.
(EI Index)
[ pdf
| ppt ]
- Li Xiang, Da Zhou and Xiaofeng Meng: A New Dynamic Hash Index for Flashbased Storage. In Proceedings of 9th International Conference on Web-Age Information Management (WAIM 2008) , Zhangjajie, China.
(EI Index)
[ pdf
| ppt ]
- Participating in ACM Conference on Information and Knowledge Management (CIKM), 2009
- Participating in China National Computer Conference (CNCC), 2009
- Participating in National Database Conference (NDBC), 2009
- Participating in Web Information Systems and Applications Conference (WISA), 2007, 2009
- Participating in International Conference on Data Engineering (ICDE), 2009
- Participating in International Conference on Mobile Data Management (MDM), 2008
- Participating in ACM SIGMOD Conference on Management of Data (SIGMOD), 2007
Last updated: October, 2009.
)