CSR: Small: Storage Architecture for the Next Generation of Smart Mobile Platforms

移动计算正在席卷全球。最近的研究指出，移动计算平台的永久存储是一个薄弱环节。由于功率和物理占用的考虑，闪存一直是迄今为止移动平台的首选存储技术。闪存和其他较新的存储技术，如PCM和STT-RAM，从架构和将它们集成到操作系统的角度来看，都有自己的特点。在本项目中，我们重点关注系统软硬件问题，以一体化的方式，构建面向移动平台的高性能存储架构。我们将结合测量、模拟和在现有和新兴移动平台上的实际实现，在不同级别的软件堆栈上分析移动平台上与存储相关的性能瓶颈的来源。我们将在系统软件和架构层面提出解决方案，以推进移动平台存储的最新技术，并对我们的解决方案进行详细的定量评估。该项目专注于推动移动存储技术的发展，可以产生意想不到的经济和社会影响。我们打算通过开源的努力使这个项目中的研究工件（跟踪、工作负载和软件）对更广泛的研究社区可用。尽管技术进步，但系统软件在提供预期性能方面明显落后于硬件。这个项目背后的愿景是试图通过集成的硬件/软件方法来弥合系统软件堆栈存储端的差距。

Mobile computing is sweeping the world. Recent studies are pointing to the permanent storage in mobile computing platforms being a weak link. Due to considerations of power and physical footprint, flash has been the storage technology of choice hitherto in mobile platforms. Flash and other newer storage technologies, such as PCM and STT-RAM, have their own idiosyncrasies both from the point of view of architecting them and integrating them in the operating system. In this project, we focus on the system software and hardware issues, in an integrated fashion, in the construction of high-performance storage architecture for mobile platforms. We will analyze the source of storage-related performance bottlenecks on mobile platforms at different levels of the software stack using a combination of measurement, simulation, and real implementation on existing and emerging mobile platforms. We will propose solutions both at the system software and architecture levels for advancing the state-of-the-art in storage for mobile platforms, and conduct detailed quantitative evaluation of our solutions. This project with its focus on advancing the state-of-the-art in mobile storage technology can have unexpected economic and societal impact. We intend to make the research artifacts from this project (traces, workloads, and software) available to the broader research community through open source efforts. Despite technological advances, system software has lagged significantly behind the hardware in delivering the expected performance. The vision behind this project is an attempt to bridge this gap for the storage side of the system software stack through an integrated hardware/software approach.