Linux 2.6 IO Performance Analysis, Quantification, and Optimization
نویسندگان
چکیده
This paper presents a novel taxonomy that characterizes in a structured and pragmatic manner the interrelationships and tradeoffs of the (rather complex) Linux 2.6 IO stack. The focus is on elaborating on the tools and techniques available in Linux 2.6 to analyze, quantify, and optimize workload-dependent IO performance. The argument made is that only a detailed, layered analysis of the Linux 2.6 application, file system, block IO layer, IO scheduler, and device driver IO chain allows optimizing the application workload onto the logical and physical IO resources. Further, the study proposes a new algorithm for the Linux 2.6 default IO scheduler CFQ. The algorithm is based on Hopfield Artificial Neural Networks (ANN) and addresses some of the potential queue and starvation issues found in the current CFQ implementation.
منابع مشابه
Workload Dependent Performance Evaluation of the Btrfs and ZFS Filesystems
The UNIX and Linux operating systems alike already provide a vast number of filesystems to the user community. In general, one of the major IO challenges faced today is scalability, especially in the lights of the very large storage IO subsystems being utilized by most IT data centers. Filesystems have to scale in their ability to address, and efficiently manage large IO storage subsystems, as ...
متن کاملSmartCon: Smart Context Switching for Fast Storage IO Devices
Handling of storage IO in modern Operating Systems assumes that such devices are slow and CPU cycles are valuable. Consequently, to effectively exploiting the underlying hardware resources, e.g. CPU cycles, storage bandwidth and etc., whenever an IO request is issued to such device, the requesting thread is switched out in favor of another thread that may be ready to execute. Recent advances in...
متن کاملDesign and Optimization of Input-Output Block using Graphene Nano-ribbon Transistors
In the electronics industry, scaling and optimization is final goal. But, according to ITRS predictions, silicon as basic material for semiconductors, is facing physical limitation and approaching the end of the path. Therefore, researchers are looking for the silicon replacement. Until now, carbon and its allotrope, graphene, look to be viable candidates. Among different circuits, IO block is ...
متن کاملMPI-IO: A Standard, Portable API for High-Performance Parallel I/O
MPI-IO, the I/O part of the MPI-2 standard, is a portable API for high-performance parallel I/O. It is speci cally designed to overcome the performance and portability limitations of the Unix-like APIs currently supported by most parallel le systems. We discuss the main features of MPI-IO and describe our MPI-IO implementation, ROMIO, which runs on most machines and le systems, including Linux ...
متن کاملAn IO Scheduling Algorithm to Improve Performance of Flash-Based Solid State Disks
Since the emergence of solid state devices into the storage scene, improvements in capacity and price have brought them to the point where they are becoming a viable alternative to traditional magnetic storage media. Current file systems and device-level I/O schedulers are optimized for rotational magnetic hard disk drives. In order to improve the efficiency of hard disk utilization, an Operati...
متن کامل