Replacing a Computer Cluster for Big Biological Data

实验技术的最新进展推动了神经科学的革命性和变革性发现。这些新技术的一个关键因素是它们产生了大量的数据。分析这些“大数据”需要非常强大的计算资源，即计算机集群。2010年，莱斯布里奇大学（UofL）加拿大行为神经科学中心（CCBN）的研究人员获得了他们的第一个计算机集群。该小组得到了广泛利用，为他们的研究做出了重大贡献。现在，在收购现有群集5年后，处理器已过时，制造商的保修已过期。预计这种高性能计算机在广泛使用5年后随时都有可能出现故障。但是，由于不切实际的高续订成本，无法延长当前群集的保修期。如果当前集群停止工作（预计在不久的将来会发生），将严重损害CCBN的生产力。因此，我们要求建立一个新的计算机集群，以取代现有的计算机集群。新的计算机集群由一个头节点和八个计算节点组成。头节点管理整个系统，并将提交的作业分发到计算节点。计算节点配备了总计约1.3 TB的内存用于大数据分析，以及约20，000个图形核心用于高度并行计算。我们还要求MATLAB软件许可证，因为MATLAB已成为神经科学和健康科学的标准分析工具。新的计算机集群将用于约25名教授在CCBN和相关部门以及约300名学员。它将支持超过50个研究项目的数据分析，这些研究项目利用尖端的电生理学、成像、细胞、分子和认知神经科学技术。此外，新的集群还将为学生提供一个获得高性能计算经验的绝佳机会。所有这些活动将大大有助于在UofL和相关机构进行的转化健康研究，从而促进加拿大人民的健康。

Recent advances in experimental technology have driven revolutionary and transformative discoveries in neuroscience. One crucial factor in these new technologies is that they produce a huge amount of data. Analysis of these ‘big data’ requires very powerful computing resources, namely a computer cluster. In 2010, the researchers at the Canadian Centre for Behavioural Neuroscience (CCBN) of the University of Lethbridge (UofL) acquired their first computer cluster. The cluster was utilized extensively and contributed to their research significantly. Now, 5 years after the acquisition of the existing cluster, the processors have become obsolete and the manufacturer’s warranty has expired. It is expected that this kind of high-performance computer is likely to break down any time after 5 years of extensive use. However, the extension of the warranty of the current cluster is not possible because of an unrealistically high renewal cost. If the current cluster stops working, which is expected to happen any time in the near future, it will significantly damage the CCBN productivity. We are therefore requesting a new computer cluster that replaces the existing one. The new computer cluster consists of one head node and eight computing nodes. The head node manages the entire system and distributes the submitted jobs to the computing nodes. The computing nodes are equipped with a total of ~1.3 TB memory for big data analysis and ~20,000 graphics core for highly parallel computation. We also request MATLAB software licenses because MATLAB has become a standard analysis tool for neuroscience and health sciences. The new computer cluster will be used by ~25 professors at the CCBN and the associated departments as well as ~300 trainees. It will support the data analysis of >50 research projects that utilize the cutting-edge electrophysiological, imaging, cellular, molecular and cognitive neuroscientific technologies. In addition, the new cluster will also provide an excellent opportunity for students to gain experience with high-performance computing. All these activities will significantly contribute to translational health researches that are performed at UofL and the associated institutions and hence to the health of people in Canada.