• Journal of the Korea Society of Computer and Information
• /
• v.23 no.3
• /
• pp.1-8
• /
• 2018

In this paper, we propose a new page replacement policy using recency and frequency on hybrid main memory. The proposal has two features. First, when a page fault occurs in the main memory, the proposal allocates it to DRAM, regardless of operation types such as read or write. The page allocated by the page fault is likely to be high probability of re-reference in the near future. Our allocation can reduce the frequency of write operations in PCM. Second, if the write operations are frequently performed on pages of PCM, the pages are migrated from PCM to DRAM. Otherwise, the pages are maintained in PCM, to reduce the number of unnecessary page migrations from PCM. In our experiments, the proposal reduced the number of page migrations from PCM about 32.12% on average and reduced the number of write operations in PCM about 44.64% on average, compared to CLOCK-DWF. Moreover, the proposal reduced the energy consumption about 15.61%, and 3.04%, compared to other page replacement policies.

• Journal of Korea Spatial Information System Society
• /
• v.11 no.2
• /
• pp.79-88
• /
• 2009

The volume of memory to store real-time data stream is varied dynamically. Continuous queries processing the data stream must manage the storage volume dynamically. In previous research, according to current volume of data a general memory manager which allocates and releases memory by a page unit is researched.However, the method frequently executes page allocation and release to store data stream. Moreover, particularly delayed queries can monopolize many of pages because the method directly allocates pages when a query has not enough memory. Focusing on the problems in memory management systems, this research proposes a memory management method which reduces the frequency of allocation and release and uniformly distributes pages for queries. The method can reduce the frequency of allocation and release through allocation based on utilization ratio of pages in each query and prevent memory monopoly through memory allocation which considers query delay.

• Journal of KIISE:Computer Systems and Theory
• /
• v.27 no.7
• /
• pp.639-649
• /
• 2000

In the relational database systems, the join operation is one of the most time-consuming query operations. Many parallel join algorithms have been developed to reduce the execution time. Multiple hash join algorithm using allocation tree is one of most efficient ones. However, it may have some delay on the processing each node of allocation tree, which is occurred in tuple-probing phase by the difference between one page reading time of outer relation and the processing time of already read one. In this paper, to solve the performance degrading problem by the delay, we develop a join algorithm using the concept of 'synchronization of page execution time' for multiple hash joins. We reduce the processing time of each nodes in the allocation tree and improve the total system performance. In addition, we analyze the performance by building the analytical cost model and verify the validity of it by various performance comparison with previous method.

• The KIPS Transactions:PartA
• /
• v.13A no.5 s.102
• /
• pp.387-398
• /
• 2006

In this paper, we propose a new page allocation scheme for flash memory file system. The proposed scheme allocates pages by exploiting the concept of Purity, which is defined as the fraction of blocks where valid Pages and invalid Pages are coexisted. The Pity determines the cost of block cleaning, that is, the portion of pages to be copied and blocks to be erased for block cleaning. To enhance the purity, the scheme classifies hot-modified data and cold-modified data and allocates them into different blocks. The hot/cold classification is based on both static properties such as attribute of data and dynamic properties such as the frequency of modifications. We have implemented the proposed scheme in YAFFS and evaluated its performance on the embedded board equipped with 400MHz XScale CPU, 64MB SDRAM, and 64MB NAND flash memory. Performance measurements have shown that the proposed scheme can reduce block cleaning time by up to 15.4 seconds with an average of 7.8 seconds compared to the typical YAFFS. Also, the enhancement becomes bigger as the utilization of flash memory increases.

• Journal of KIISE:Computer Systems and Theory
• /
• v.37 no.5
• /
• pp.272-284
• /
• 2010

Flash Memory differs from the hard disk, because it cannot be overwritten. Most of the flash memory file systems use not-in-place update mechanisms for the update. Flash memory file systems execute sometimes block cleaning process in order to make writable space while performing not-in-place update process. Block cleaning process collects the invalid pages and convert them into the free pages. Block cleaning process is a factor that affects directly on the performance of the flash memory. Thus this paper suggests the efficient page allocation method, which reduces block cleaning cost by minimizing the numbers of block that has valid and invalid pages at a time. The result of the simulation shows an increase in efficiency by reducing more block cleaning costs than the original YAFFS.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.13 no.3
• /
• pp.109-116
• /
• 2018

SSDs that are not allowed in-place update within the allocated page cause another allocation of a new page that will replace the previous page at the moment data modification occurs. This intrinsic characteristic of SSDs requires many changes to the existing HDD-based IO theory. In this paper, we conduct a performance comparison of FTL caching strategy in perspective of cache hashing (Global vs. grouped) and caching algorithm (LRU vs. NUR) through a simulation. Experimental results show that in terms of energy consumption for flash operation the grouped management of cache is not suitable and NUR algorithm is superior to LRU algorithm. In particular, we found that the cache hit ratio of LRU algorithm is about 10% point higher than that of NUR algorithm while the energy consumption of LRU algorithm is about 32% high.

• Journal of the Korea Society of Computer and Information
• /
• v.21 no.2
• /
• pp.1-7
• /
• 2016

In this paper, we propose in this paper present a novel locality data allocation policy as COLD(Correlated Locality Data) allocation policy. COLD is defined as a set of data that will be updated together later. By distributing a COLD into a NAND block separately, it can preserve th locality. In addition, by handling multiple COLD simultaneously, it can obtain the parallelism among NAND chips. We perform two experiment to demonstrate the effectiveness of the COLD data allocation policy. First, we implement COLD detector, and then, analyze a well-known workload. And we confirm the amount of COLD found depending on the size of data constituting the COLD. Secondly, we compared the traditional page-level mapping policy and COLD for garbage collection overhead in actual development board Cosmos OpenSSD. Experimental results have shown that COLD data allocation policy is significantly reduces the garbage collection overhead. Also, we confirmed that garbage collection overhead vary depending on the COLD size.

• Journal of KIISE:Databases
• /
• v.28 no.4
• /
• pp.732-741
• /
• 2001

In the relational database systems the join operation is one of the most time-consuming query operations. Many parallel join algorithms have been developed 개 reduce the execution time Multiple hash join algorithm using allocation tree is one of the most efficient ones. However, it may have some delay on the processing each node of allocation tree, which is occurred in tuple-probing phase by the difference between one page reading time of outer relation and the processing time of already read one. This delay problem was solved by using the concept of synchronization of page execution time with we had proposed In this paper the effects of the performance improvements in each node of the allocation tree are extended to the whole allocation tree and the performance evaluation about that is processed. In addition we propose an efficient algorithm for multiple hash joins in limited number of processor environments according to the relationship between the number of input relations in the allocation tree and the number of processors allocated to the tree. Finally. we analyze the performance by building the analytical cost model and verify the validity of it by various performance comparison with previous method.

• Journal of KIISE:Computer Systems and Theory
• /
• v.34 no.11
• /
• pp.599-609
• /
• 2007

Due to the restrictions of Flash memory such as overwrite limitation and write/erase operational unit differences, block cleaning is required in Flash memory based file systems and known as a key factor on the performance of file systems. In this paper, we identify three parameters, namely utilization, invalidity and uniformity, and analyze how the parameters affect the cost of block cleaning. The analysis show that as uniformity degrades, the cost of block cleaning increases drastically. To overcome this problem, we design a new modification-aware(MODA) page allocation scheme that strives to keep uniformity high by separating frequently-updating data from infrequently-updating data. Real implementation experiments conducted on an embedded system show that the MODA scheme can actually enhance uniformity of Flash memory, which consequently leads to reduce the cost of block cleaning with an average of 123%, compared to the traditional sequential allocation scheme that is used in YAFFS.

• Proceedings of the Korean Information Science Society Conference
• /
• 2005.11b
• /
• pp.103-105
• /
• 2005

현재 UNISYS 네트워크 데이터베이스(Network Data Base, NDB)에서는 엔트리 레코드(entry record)를 배치시키기 위한 방법 중 CALC방식의 DMSCALC 알고리즘을 사용하고 있다. 그러나 이 알고리즘의 특성상 하위 레코드가 과다하게 발생하는 업무의 경우 동일 기본저장 페이지(primary page)에 할당되는 타 엔트리 레코드의 저장 영역에 대한 제한을 유발시켜 오우버플로워 페이지(overflow page)로 이동되고 이에 따라 성능 저하를 가져온다. 본 논문에서는 나눗셈법을 따르는 DMSCALC 알고리즘을 분석 연구하고, 엔트리 레코드가 저장되는 기본저장 페이지의 산출 알고리즘을 개선하여 기본저장 페이지 간 간격을 확보하였으며 개선 전 알고리즘과 비교할 때 기본저장 페이지 사용 율의 경우 최소 $22\%$, 최대 $41\%$에 이르는 성능 향상을 나타내었다. 또한 오우버플로워 페이지 발생률의 경우에도 최소 $47\%$, 최대 $67\%$의 감소 효과를 나타내었다.