CSR: Small: Efficient Usage of Shingled Magnetic Recording (SMR)Drives

过去十年见证了计算、有线和无线通信以及存储技术的巨大进步。同样重要的是，显著的成本降低使大容量的计算和存储能力日益可用。许多网络被集成到称为互联网的世界上最大的IT(信息技术)基础设施中。随着互联网提供前所未有的连通性，海量的数字数据被创造并提供给人们。存储正在生成的海量数据已成为一项重大挑战。与此同时，更多的数据类型，如音频和视频，被转换为数字格式，各种类型的数据需要保存很长时间(超过100年)。因此，我们确实处于一个大数据时代。大数据问题目前不仅是个人面临的问题，而且几乎是社会所有阶层都面临的问题，包括专业团体、研究社区、公司和政府。在这个解决存储和访问大数据挑战的项目中，计划研究一种新的高效方法，使用一种名为叠片磁记录(SMR)的有前途的技术在磁盘驱动器(称为叠片写入磁盘)中存储大量数据。基于SMR的驱动器是一种很有前途的技术，可能会满足大数据的存储需求。由于SMR的好处来自写入期间的重叠磁道，因此在执行数据更新时研究新的用途以减少潜在的写入放大是很重要的。该项目的计划是开发可减少写入放大的新颖布局设计和专用文件系统，使应用程序无需任何修改即可使用此类设备，以及开发在SMR驱动器上高效支持基于树和图形的数据结构以用于许多大数据相关应用的新方法。这项工作的顺利完成将提供一种满足大数据需求的新型存储设备。

The past decade has witnessed tremendous advances in computing, wired and wireless communication and storage technologies. It is also important that remarkable cost reductions have made large computing and storage capacity increasingly available. Many networks are integrated into the world's largest IT (Information Technology) infrastructure called the Internet. With the unprecedented connectivity provided by Internet, a huge amount digital data has been created and made available. Storage for the huge amount of data being generated has become a significant challenge. At the same time, more data types like audio and video are converted into digital format and data in various types needs to be preserved for a very long term (beyond 100 years). Thus, we are indeed in a big data era. The big data problem is currently faced by not only individuals, but also by almost all segments of the society, including professional groups, research communities, companies and government. In this project to address the challenges of storing and accessing big data, the plan is to investigate new and efficient ways of using a promising technology called Shingled Magnetic Recording (SMR) for storing large volumes of data in magnetic disk drives (called Shingled Write Disks). SMR based drives are a promising technology that may answer the storage demand of big data. Since the benefits of SMR come from overlapping tracks during writes, it is important to investigate novel uses to reduce the potential write amplification when updates of data are performed. The project plan is to develop both novel layout designs that can reduce write amplification and specialized file systems that can enable applications to use such a device without any modifications, as well as, novel ways of efficiently supporting tree and graph based data structures on SMR drives for many big data related applications. The successful completion of this work will provide a new type of storage devices meeting the demand of big data.