关系数据库缓冲区置换策略的性能评测(英文)

缓冲区是计算机领域一个非常重要的研究主题。现在存在很多缓冲区置换策略,广泛地用于操作系统、文件系统、数据库以及存储系统中。提出了一种性能评测方法来评测关系数据库的缓冲区置换策略。这种方法采用直接测试方式,使用五个性能指标,包括新提出的指标以及由于模拟方法的局限性在以前的研究中很少使用的性能指标,同时利用显著性测试并提出一种联合使用多个显著性测试的方法进行置换策略之间的比较,设计和使用多种工作负载在单用户和多用户情况下有效地评测各种策略。使用这种评测方法,在一个真实的关系数据库环境中,对十三种典型的缓冲区置换策略的性能进行了详细、全面的实验研究,从中可以获得一些有用的结论。     

BENCHMARKING RIVER HABITAT IMPROVEMENT

River ecosystems have witnessed a long history of human pressure, particularly the disruption of freshwater fish populations. The awareness of this situation has led to many habitat improvement projects, with a variable degree of success. In natural situations, fish populations co‐inhabit throughout the hydrological cycle with different degrees of adequacy, and the sequence of favourable and unfavourable conditions dictates abiotic constraints and biotic interactions that shape the final biological assemblages. We postulate that a part of unsuccessful restoration results is related to insufficient closeness to the natural habitat conditions of the river type that is to be restored, including the naturally adverse periods. We used the river2d model to predict habitat availability as weighted usable area (WUA) at a degraded site that is to be restored, for two native Mediterranean species and their life stages—the Southwestern nase Iberochondrostoma almacai and the Arade chub Squalius aradensis. We then analysed the yearly evolution of the natural WUA at a nearby reference site. Overall, the reference site exhibited the longest periods during which the WUA was continuously lower than the chosen WUA thresholds for each of the four bioperiods. Considerable divergences from natural habitat availability values can be seen for the spawning, rearing and growth bioperiods. Restoration outcomes can result in appreciable deviations—favourable or unfavourable to fish populations—from the WUA occurring under natural conditions over the course of the year. Restoration should therefore take account of local hydraulic and habitat patterns that govern population dynamics and result in the final fish assemblage. Copyright © 2011 John Wiley & Sons, Ltd.     

在运动参考系内如何正确测量时间膨胀

讨论并论证了如何正确测量时间膨胀的方法,指出了在相对于新以太运动的参考系内人们只能正确测量往返运动钟发生的时间膨胀,但却不能正确测量单向运动钟发生的时间膨胀,并以此解释了狭义相对论难以解释的一些现象,消除了人们在分析实验时所出现的多重标准.     

基于线性预测的移动终端LPDVFS策略

针对以Linux为内核的移动操作系统,提出一种细粒度的DVFS策略LPDVFS。该策略基于历史数据,使用线性预测的方法,指导电压频率调整方向和幅度。线性预测中的参数通过回归方法确定。实验结果表明,LPDVFS策略相比Linux内核默认使用的粗粒度调频策略,能降低系统13.55%的功耗,延长移动终端的续航时间。     

在线事务处理系统性能评测框架研究

在线事务处理(online transaction processing,OLTP)系统在不同的应用环境下存在着模型的差异,但是作为典型的数据库应用系统,其在不同应用下的性能评测又存在着一定的共性.针对用户自身应用领域的性能评测环境,提出了一种针对OLTP系统进行性能评测的通用框架,用户可以动态配置数据规模、事务混合比例等测试参数.在框架基础上,以银行中间业务这一典型的OLTP应用为例,实现了面向银行中间业务的OLTP性能评测系统,通过实验验证了评测系统及所提框架的有效性.     

分布式文件存储系统性能测试问题研究

分布式文件存储系统的性能测试不同于普通的文件系统性能测试。首先,需要根据系统支持的操作确定测试的benchmark,然后,需要根据系统的设计规模对多用户并发访问系统的情形进行模拟。在“燕星”分布式文件存储系统的性能测试中,对Andrewbenchmark进行改造,确定了测试的benchmark;基于RMIcallback机制,模拟了500个用户并发访问系统的情形。并将测试结果与NFS进行了对比,结果表明,“燕星”系统具有良好的实用性能。     

NPB CG在分布式环境下的并行实现

ＣＧＢｅｎｃｈｍａｒｋ是ＮＡＳＰａｒａｌｅｌＢｅｎｃｈｍａｒｋｓ（ＮＰＢ）中的一个核心程序，它用共轭梯度法求大型稀疏对称正定矩阵的最小特征值，本文介绍其主要算法，并给出在分布式环境下的高效并行算法，最后给出了在ＳＧＩＣｈａｌｅｎｇｅＰＶＭ平台上的测试结果     

网络攻击行为分类技术的研究

文章在研究国外网络攻击行为分类技术的基础上，从安全系统和安全产品评测的目标出发，提出了适合评测需要的网络攻击行为分类体系，阐述考虑分类问题着眼点以及确定的攻击分类原则和攻击分类标准。     

Use of the surface harmonics method forcalculation of 2D C5G7 MOX benchmark

The C5G7 MOX benchmark specifying a sixteen-assembly core with asurrounding water reflector was proposed as a basis to measure current transport code abilities in the treatment of reactor core problems without spatial homogenization. Seven-group cross sections for all materials were used as initial information. Just that fact allows to test an accuracy of solving the neutron transport equation excluding additional errors connected with preparing the group cross sections. In this paper, Surface Harmonics Method (SHM) is applied to calculation of the two-dimensional configuration of this benchmark. Different approximations of SHM were applied, both with and without spatial homogenization. Additionally, this fact allowed evaluating the effect of spatial homogenization of cells. Comparisons were carried out for k_eff and pin powers both with the reference results and between the results calculated by different SHM approximations.     

定位误差的本质及其计算方法讨论

定位误差的分析和计算是进行工艺装备设计必不可少的重要环节。文中根据工程实践及教学体会，分析了计算定位误差的原因和定位误差的本质，并结合具体的实例．对计算定位误差的三种方法——定义法、合成法、微分法进行了阐述，为正确选择计算方法提供了依据。