플래시메모리 SSD를 이용한 확장형 버퍼 관리

심도윤(Doyoon Sim),박장우(Jangwoo Park),김성탄(Sungtan Kim),이상원(Sangwon Lee),문봉기(Bongki Moon) 한국정보과학회 2010 정보과학회논문지 : 데이타베이스 Vol.37 No.6

최근 들어, 플래시메모리의 가격이 지속적으로 낮춰지고, 플래시메모리 기반 SSD 컨트롤러 기술이 급격하게 발전하면서 중저가의 고성능 플래시 SSD가 시장에 널리 보급되고 있다. 하지만, 데이터베이스 분야에서 가격 등의 이유로 당분간 플래시 SSD가 하드디스크를 완전히 대체하기는 쉽지 않을 것이다. 대신 플래시 SSD의 빠른 성능을 캐시 용도로 활용하는 접근법이 현실적이고, 실제로 하드디스크와 플래시메모리를 하이브리드 형태로 사용하는 접근법들이 제시되었다. 본 논문에서는 기존의 접근법들과는 달리, 플래시 SSD를 데이터베이스의 버퍼에서 밀려나는 페이지들을 순차적으로 저장하고, 재 참조될 때 하드디스크 대신 플래시 SSD에서 읽혀지도록 하는 확장 버퍼 아키텍처를 제안한다. 플래시 SSD를 저장장치 레벨에서 캐시로 사용하는 기존 방법들에 비해, 플래시 SSD를 호스트 시스템에서 확장 버퍼로 사용함으로써 읽기 측면에서 주 버퍼에서 밀려나는 웜 페이지(warm page)들에 대해 상당한 성능 개선을 이룰 수 있다. TPC-C 트레이스를 사용한 시뮬레이션 결과, 주 버퍼에 없는 페이지들이 확장 버퍼에서 찾아지는 적중률이 60%를 넘는 사실을 알 수 있었다. 이 확장 버퍼아키텍처는, 동일한 비용을 지불하는 다른 접근법, 즉 DRAM을 버퍼로 추가하는 기법과 하드디스크를 추가하는 기법에 비해 가격 대비 성능 개선 효과가 높다. As the price of flash memory continues to drop and the technology of flash SSD controller innovates, high performance flash SSDs with affordable prices flourish in the storage market. Nevertheless, it is hard to expect that flash SSDs will replace harddisks completely as database storage. Instead, the approach to use flash SSD as a cache for harddisks would be more practical, and, in fact, several hybrid storage architectures for flash memory and harddisk have been suggested in the literature. In this paper, we propose a new approach to use flash SSD as an extended buffer for main buffer in database systems, which stores the pages replaced out from main buffer and returns the pages which are re-referenced in the upper buffer layer, improving the system performance drastically. In contrast to the existing approaches to use flash SSD as a cache in the lower storage layer, our approach, which uses flash SSD as an extended buffer in the upper host, can provide fast random read speed for the warm pages which are being replaced out from the limited main buffer. In fact, for all the pages which are missing from the main buffer in a real TPC-C trace, the hit ratio in the extended buffer could be more than 60%, and this supports our conjecture that our simple extended buffer approach could be very effective as a cache. In terms of performance/price, our extended buffer architecture outperforms two other alternative approaches with the same cost, 1) large main buffer and 2) more harddisks.

Suggestion on What to Prepare for the Performance Assessment for the Deep Geological Facility in Korea: Focusing on Buffer Mass Loss

Gitae Kim,Seong-Su Jeon,Sukhoon Kim,Jeong-Hwan Lee,Haeryong Jung 한국방사성폐기물학회 2023 한국방사성폐기물학회 학술논문요약집 Vol.21 No.1

In buffer, a main component of engineering barrier system (EBS) in the deep geological repository, mass loss is mainly caused by upheave and mechanical erosion. The former is a phenomenon that bentonite in the upper part of the buffer moves to the backfill region due to groundwater intake and swelling. And, the latter is a phenomenon that bentonite on the surface of the buffer moves to the backfill region due to groundwater flow at the interface with host rock as the buffer saturates. Buffer mass loss adversely affects the fulfilment of the safety function of the buffer that is to limit and retard radionuclide release in the event of canister failure. Accordingly, in this paper, we reviewed how to consider this phenomenon in the performance assessment for the operating license application in Finland, and tentatively summarized data required to conduct the analysis for the domestic facility based on the review results. Regarding buffer mass loss, the previous studies carried out in Finland are categorized as follows: 1) experiment on the amount of buffer upheave with groundwater inflow rate (before backfilling), 2) analysis for the amount of buffer upheave with groundwater inflow rate (after backfilling), 3) analysis of buffer erosion rate with groundwater inflow rate, 4) analysis for distribution of the groundwater inflow rate into the buffer for all deposition holes (using ConnectFlow modeling results), and 5) analysis of buffer mass loss with groundwater salinity. Finally, the buffer mass loss distribution table was derived from the results of 1) through 3) by combining with that of 4). Given these studies, the following will be required for the performance assessment for buffer mass loss in the domestic disposal facility: a) distribution table of buffer mass loss for combined interactions taking into account effect of 5) (i.e. 1), 2), 3), and 5) + 4)), and b) Threshold for buffer mass loss starting to negatively affect the fulfilment of the safety function of the buffer. Even though it is judged that the results of this study could be directly applied to developing the design concept of EBS and to conducting the performance assessment in the domestic disposal facility, it is essential to prepare a set of input data reflecting the site-specific design features (e.g. dimension, material used, site, etc.), which include saturation time and groundwater salinity.

Design and Analysis of a Buffer Controller Using Buffer Locality

Kim, Wankyoo,Park, Soheui,Moon, Beongcheol 진주산업대학교 산업과학기술연구소 1996 산업과학기술연구소보 Vol.- No.3

In multimedia communication systems, the media stream units are produced at a regular time interval. As the packets pass through the network, each can encounter a varying amount of queueing delay. Therefore it is difficult to maintain the regular time interval. This situation leads the receiver's buffer state to starvation or to overrun, so the continuous properties of multimedia streams will not be guaranteed. In order to solve these problems, we design a buffer controller and analyze it in this paper. The critical sections of overrun and starvation are established in the receiver's buffer. Whenever the periodic BCT(Buffer Check Time) is invoked, the buffer controller inspects the state of current buffer and calculates the sizes of region buffer and transition buffer, it can control the magnitude of media scaling policy based on these results. The buffer controller guarantees the quality of the media stream, increasing the utility of the rereiver's resource and makes the fast response of media scaling using the properties of buffer locality.

메모리 자원 사용 효율성 증진을 위한 적응적 네트워크 이중 버퍼 모델

최창범(Daniel Choi),이승룡(Sung Young Lee) 한국정보과학회 2006 정보과학회논문지 : 시스템 및 이론 Vol.33 No.11·12

본 논문에서는 네트워크 통신에서 혼잡으로 인한 패킷의 손실을 최소화하기 위하여 새로운 버퍼 모델인 적응적인 이중 버퍼 모델을 제안한다. 이는 제약된 메모리 환경에서 송수신 버퍼가 서로의 여유 공간을 공유하여 패킷의 손실을 최대한 줄일 수 있는 버퍼 모델이다. 또한 리스트와 비슷한 성능을 지니는 본 버퍼 모델은 자유 리스트를 사용한 버퍼와 달리 메모리 누수로 인한 버블(bubbles) 현상을 방지하므로 제한된 환경의 네트워크 버퍼에 적용할 수 있으며 배열을 사용하는 경우와 비교 할 때 최대 100%성능 향상을 기대할 수 있다. This paper proposes an Adaptive Double Buffer Model. As a new FIFO buffer model, this technique minimizes packet losses from network congestion by logically managing buffers. It allocates the spare spaces of non-congested buffers to congested buffers by allowing receive/send buffers to share two queues, and hence it minimizes packet losses. In contrast to the buffer model utilizing a free list, this buffer model can prevent the bubble phenomenon caused by a memory leak and thereby apply to a network buffer in a restricted environment. Also, compared with the model using an array, this model brings maximum 100 percent improvement in accepting packets and compared with the model utilizing a free list, this model has the similar efficiency. Results of the performance test on Adaptive Double Buffer Model, shows that this proposed model decreases packet losses and enhances memory efficiency.

가시나무류의 RAPD 유전분석을 위한 Genomic DNA 최적 추출조건 및 RAPD Primer 선별

박동진,최명석 경상대학교 농업생명과학연구원 2018 농업생명과학연구 Vol.52 No.2

유전분석을 위해서, CTAB buffer를 이용하여 가시나무류 4종의 genomic DNA를 분리하였다. CTAB buffer를 이용하여 분리한 genomic DNA의 순도는 Edward buffer를 이용했을 때보다 더 높게 나타났다. 가시나무(Q. myrsinaefolia)로부터적정 순도 gDNA는 2% CTAB에 0, 1 또는 2% PVP가 첨가된 buffer를 이용했을 때 얻어졌다. 종가시나무(Q. glauca)로부터 1% CTAB buffer와 1% PVP를 첨가한 buffer를 이 용하여 얻은 gDNA 순도는 1.84±0.02(A260/280)였다. 참가시나무(Q. salicina)의 적정순도 gDNA는 2% CTAB와 1 또는 5% PVP가 첨가된 buffer를 이용하여 분리할 수 있었다. 2% CTAB와 2% PVP가 첨가된 buffer를 이용했을 때, 졸가시나무(Q. phillyraeoides)의 gDNA의 순도는 1.85±0.01(A260/280)였다. 18개 의 RAPD primer를 사용하여 PCR을 수행하였을 때, 가시나무에서는 15개 primer에 의해 53개의 다형질 DNA가 발견되었다. 종가시나무에서는 11개의 primer에 의해 40개의 다형질 DNA가 나타났다. 참가시나무의 경우 16개의 primer에 의해 50개의 증폭된 DNA밴드를 확인 할 수 있었다. 졸가시나무에서 14의 primer가 증폭반응을 일으켰고, 53개의 다형질 DNA가 나타났다. 이 결과들은 가시나무류의 유전분석에 이용할 수 있다. For genetic analysis of four species of evergreen Quercus spp, their genomic DNA were isolated using CTAB buffer method. The purity values of genomic DNA of each species isolated using CTAB buffer were higher than those isolated using Edward buffer. The adequate purity of genomic DNA of Quercus myrsinaefolia was obtained using the buffer containing 2% CTAB and 0, 1 or 2% PVP. The purity of gnomic DNA of Q. glauca was 1.84±0.02(A260/280) when the buffer containing 1% CTAB and 1% PVP was used. Adequate genomic DNA purity of Q. salicina could be isolated using the buffer containing 2% CTAB and 1 or 5% PVP. The purity of genomic DNA of Q. phillyraeoides was 1.85±0.01(A260/280), and it was obtained when the buffer 2% CTAB and 2% PVP was used. When the PCR reactions using 18 RAPD primers were performed, 15 primer produced 53 amplified fragments of Q. myrsinaefolia. Eleven primers produced 40 polymorphic DNA of Q. glauca. For amplification of random polymorphic DNA of Q. salicina, 16 primers produced 50 bands. Of Q. phillyraeoides, 53 polymorphic bands were amplified by 14 random primers. These results can be used for genetic analysis of evergreen Quercus spp.

1/<i>f</i> noise characteristics of AlGaN/GaN HEMTs with periodically carbon-doped GaN buffer layer

Im, Ki-Sik,Choi, Jinseok,Hwang, Youngmin,An, Sung Jin,Roh, Jea-Seung,Kang, Seung-Hyeon,Lee, Jun-Hyeok,Lee, Jung-Hee Elsevier 2019 MICROELECTRONIC ENGINEERING Vol.215 No.-

<P><B>Abstract</B></P> <P>We investigate the DC and 1/<I>f</I> noise properties in Al<SUB>0.25</SUB>Ga<SUB>0.75</SUB>N/GaN high-electron mobility transistors (HEMTs) with two types of 2 μm-thick periodically carbon-doped GaN buffer layer (PC-doped GaN buffer) with and without inserting the 30 nm-thick Al<SUB>0.05</SUB>Ga<SUB>0.95</SUB>N back barrier layer between the GaN channel layer and the PC-doped GaN buffer. The PC-doped GaN buffer layer consists of multiple layers of 12 nm-thick C-doped GaN layer with doping concentration of 1 × 10<SUP>18</SUP> cm<SUP>−3</SUP> and 50 nm-thick undoped GaN layer with unintentional n-typing concentration of 2 × 10<SUP>16</SUP> cm<SUP>−3</SUP>. A reference AlGaN/GaN HEMT with 2 μm-thick highly-resistive GaN buffer layer without C-doping is also fabricated for comparison. Similarly to the reference AlGaN/GaN HEMT, the AlGaN/GaN HEMTs with PC-doped GaN buffer show typical 1/<I>f</I> noise characteristics mainly due to the trapping effects at the AlGaN/GaN interface from subthreshold region to strong-accumulation region, which indicates that the deep trapping effects in the PC-doped GaN buffer layer is negligible, and experience the correlated mobility fluctuations (CMF), which is convinced from the drain current power spectral density (PSD) versus drain current. At off-state (deep-subthreshold region), on the other hand, the HEMTs with the PC-doped GaN buffer layer exhibit 1/<I>f</I> <SUP>2</SUP> noise characteristics, which are closely related to the generation-recombination (g-r) noise caused by the spatial trapping/detrapping process between the deep acceptor in the C-doped layer and the shallow donor in the undoped layer in the PC-doped GaN buffer, while the reference HEMT still shows typical 1/<I>f</I> noise characteristics.</P> <P><B>Highlights</B></P> <P> <UL> <LI> Noise characteristics in AlGaN/GaN HEMTs with/without PC-doped buffer layer were investigated. </LI> <LI> PC-doped buffer layer consists of 12 nm-thick carbon-doped GaN and 50 nm-thick un-doped GaN. </LI> <LI> All devices exhibited 1/<I>f</I> noise properties and CMFs from subthreshold to strong-accumulation. </LI> <LI> At off-state, PC-doped buffer devices exhibited 1/<I>f</I> <SUP>2</SUP> noise properties at frequency > 40 Hz. </LI> </UL> </P> <P><B>Graphical abstract</B></P> <P>[DISPLAY OMISSION]</P>

Study of Buffer Substitution Using Inhibitory Compound CaCO₃ in Enzymatic Hydrolysis of Paper Mill Waste Fines

Byeongcheol Min,Surya N. Jampana,Christoper M. Thomas,Bandaru V. Ramarao 한국펄프·종이공학회 2018 펄프.종이技術 Vol.50 No.2

Buffer substitution method were studied to improve the efficiency and to reduce cost of enzymatic hydrolysis of paper mill waste fines. The method is utilization of CaCO₃ in waste fine as a buffer material to substitute traditional buffer chemicals using in hydrolysis process. The 0.05M of sodium acetate buffer required 8.2 ㎏ of sodium acetate and 2.0 ㎏ of acetic acid per 100 ㎏ of waste fines to make the pH 5. When buffer solution and waste fines were mixed, the pH of the mixture shifted to 6.5-7.0 due to the dissociation of CaCO₃. The shifted pH was adjusted with addition of extra acids such as sulfuric acid and acetic acid to reach the pH to 5.2. Adjustment of pH increased waste fine hydrolysis yield from 33.7% to 41.4% by addition of extra acetic acid as much as 6.3 ㎏ per 100 kg of waste fines. For the buffer substitution method, only 7.7 ㎏ of acetic acid was demanded without sodium acetate for the same pH level (pH 5.2) of the mixture. The buffer substitution method resulted in 40.8% of hydrolysis yield which was comparable result of pH adjustment of buffer. The buffer substitution method reduced the requirement of 8.2 ㎏ of sodium acetate and 0.6 ㎏ of acetic acid for 100 ㎏ of waste fine hydrolysis. Addition of sulfuric acid resulted adverse effects compared to addition of acetic acid for both pH adjustment and buffer substitution.

The Relationship between Capital Buffer and Contemporarily Financial Policy under Economic Condition: Evidence from Korean Credit Card Firms

서지용 한국자료분석학회 2013 Journal of the Korean Data Analysis Society Vol.15 No.6

This study examines the relationship between capital buffer of Korean credit card firms and financial policy under economic condition such as business cycle and financial market situation. The main focus of this study is to review whether Korean credit card firms can meet the capital buffer requirement pursuing by supervisory authorities under regulation for leverage and fee limitations in terms of precautionary motivation perspectives. According to the analysis results, first, business cycle did have positively significant influence on capital buffer. Thus, the pro-cyclicality of capital buffer to business cycle exists in Korean credit card firms. Second, market rate and foreign exchange rate have significantly impact on capital buffer. Due to increasing the cost of capital for credit card firms, the relationship between capital buffer and market rate or foreign exchange rate is negative. Specifically, higher interest rates and weaker Korean currency reduce the amount of capital buffer. Third, leverage limit of supervisory authorities leads to increased capital buffer as suggested by Milne, Whalley (2001). However, it is found that tight fee regulation does not contribute to increasing capital buffer. As the policy implication, stronger leverage limit of supervisory authorities results in making the Korean credit card firms reserve up more capital buffer.

NAND 플래시 메모리에서 쓰기/지우기 연산을 줄이기위한 버퍼 관리 시스템

정보성(Bo-Sung Jung),이정훈(Jung-Hoon Lee) 한국컴퓨터정보학회 2011 韓國컴퓨터情報學會論文誌 Vol.16 No.10

NAND 플래시 메모리는 저전력, 저렴한 가격, 그리고 대용량임에도 불구하고 페이지 단위의 쓰기 및 블록 단위의 지우기 연산은 큰 문제점을 가지고 있다. 특히 NAND 플래시 메모리 특성상 덮어쓰기가 불가능하므로 쓰기동작 후 수반되는 지우기 동작은 전체 성능저하의 원인이 된다. 기존의 NAND 플래시 메모리를 위한 SRAM 버퍼는 간단하면서도 NAND 플래시 메모리의 쓰기 동작을 효과적으로 줄여줄 수 있을 뿐 아니라 빠른 접근 시간을 보장 할 수 있다. 본 논문에서는 작은 용량의 SRAM을 이용하여 NAND 플래시 메모리의 가장 큰 오버헤드인 지우기/쓰기 동작을 효과적으로 줄일 수 있는 버퍼 관리 시스템을 제안한다. 제안된 버퍼는 큰 페칭 크기를 가지는 공간적 버퍼와 작은 페칭 크기를 가지는 시간적 버퍼인 완전연관 버퍼로 구성된다. 시간적 버퍼는 공간적 버퍼에서 참조된작은 페칭을 가지며, NAND 플래시 메모리에서 쓰기 및 지우기 수행시 시간적 버퍼내에 존재하는 같은 페이지 혹은 블록에 포함된 페칭 블록을 찾아 동시에 처리한다. 따라서 NAND 플래시 메모리에서 쓰기 및 지우기 동작을 획기적으로 줄였다. 시뮬레이션 결과에 따르면 제안된 NAND 플래시 메모리 버퍼 시스템은 2배 크기의 완전연관 버퍼에 비해 접근 실패율 관점에서는 높았지만, 쓰기 동작과 지우기 동작은 평균적으로 각각 58%, 83% 정도를 줄였으며, 결론적으로 평균 플래시 메모리 접근 시간은 약 84%의 성능 향상을 이루었다. There are the large overhead of block erase and page write operations in NAND flash memory, though it has low power consumption, cheap prices and a large storage. Due to the physical characteristics of NAND flash memory, overwrite operations are not permitted at the same location, so rewriting operation require after erase operation. it cause performance decrease of NAND flash memory. Using SRAM buffer in traditional NAND flash memory, it can not only reduce effective write operation but also guarantee fast memory access time. In this paper, we proposed the small SRAM buffer management system for reducing overhead of NAND flash memory, that is, erase and write operations. The proposed buffer system in a NAND flash memory consists of two parts, i.e., a fully associative temporal buffer with the small fetching block size and a fully associative spatial buffer with the large fetching block size. The temporal buffer have small fetching blocks that referenced from spatial buffer. When it happen write operations or erase operations in NAND flash memory, the related fetching blocks in temporal buffer include a page or a block are written in NAND flash memory at the same time. The writing and erasing counts in NAND flash memory can be reduced. According to the simulation results, although we have high miss ratios, write and erase operations can be reduced approximatively 58% and 83% respectively. Also the average memory access times are improved about 84% compared with the fully associative buffer with two sizes.

제한된 수의 튜닝 가능한 파장변환기와 내부파장을 갖는 하이브리드 버퍼 구조의 광 패킷 스위치

임헌국 ( Huhnkuk Lim ) 한국인터넷정보학회 2009 인터넷정보학회논문지 Vol.10 No.2

광 패킷 스위칭 기술은 대용량 인터넷 트래픽을 패킷레벨에서 보다 유동적인 대역폭으로 전송해 줄 수 있으며, WDM 계층과 I P 계층을 유연하게 통합 해 줄 수 있기 때문에 차세대 광 인터넷을 위한 가장 강력한 스위칭 기술 중 하나이다. 광 패킷 스위칭 기술의 연구는 동기식 망 안에서의 연구와 비동기식 망 안에서의 연구로 나뉘며 본 논문에서는 인터넷 트래픽과 같은 비동기 가변길이 패킷들을 위한 광 패킷 스위치 내부에 경합해결 이슈에 초점이 맞추어져 있다. 하이브리드 버퍼는 광 지연 선로 버퍼와 전자 버퍼로 구성 되어 비동기 가변길이 패킷들의 경합 해결을 위한 하나의 대체 버퍼로서 패킷손실 확률을 감소시키기 위해 소개된바 있다. 그런데 스위치 내부의 자원 중 튜닝 가능한 파장 변환기의 개수와 내부 파장개수를 고려한 광 패킷 스위치의 디자인은 스위치 비용과 자원 이용 효율을 결정짓는 매우 중대한 이슈이다. 따라서 본 논문에서는 한정된 수의 튜닝 가능한 파장 변환기와 내부 파장 개수를 고려한 하이브리드 버퍼 구조형 광 패킷 스위치를 설계하고 그것의 스케쥴링 알고리즘을 소개하고자 한다. 제안한 스케쥴링 알고리즘은 광 지연 선로 버퍼만을 갖는 기존의 LAUC-VF알고리즘과 비교하여 더 나은 패킷 손실 확률을 제공해 줄 수 있었다. Optical packet switching (OPS) is a strong candidate for the next-generation internet, since it has a fine switching granularity at the packet level for providing flexible bandwidth, and provides seamless integration between WDM layer and IP layer. Optical packet switching have been studied in two categories: OPS in synchronous and OPS in asynchronous networks. In this article we are focused on contention resolution of OPS in asynchronous networks. The hybrid buffer have been addressed, to reduce packet loss further as one of the alternative buffer structures for contention resolution of asynchronous and variable length packets, which consists of the FDL buffer and the electronic buffer. The OPS design issue for the limited number of TWCs and internal wavelengths is important in the aspect of switch cost and resource efficiency. Therefore, an hybrid buffer structured optical packet switch and its scheduling algorithm is presented for considering the limited number of TWCs and internal wavelengths, for contention resolution of asynchronous and variable length packets. The proposed algorithm could lead to the packet loss improvement compared to the legacy LAUC-VF algorithm with only the FDL buffer.