Block I/O

  • | The difference (between character device and block device) comes down to | whether the device accesses data randomly — in otherwords, whether the | device can seek to one position from another.

  • | The block is an abstraction of the filesystem — filesystems can be | accessed only in multiples of a block.Although the physical device is | addressable at the sector level, the kernel performs all disk operations in | terms of blocks.

  • | the kernel (as with hardware and the sector) needs the block to be a power | of two.The kernel also requires that a block be no larger than the page size.

  • | The purpose of a buffer head is to describe this mapping between the | on-disk block and the physical in-memory buffer (which is a sequence of bytes | on a specific page).

  • | the kernel does not issue block I/O requests to the disk in the order they | are received or as soon as they are received.

  • | Both the process scheduler and the I/O scheduler virtualize a resource | among multiple objects.

  • I/O schedulers perform two primary actions to minimize seeks: merging and sorting. a) Merging is the coalescing of two or more requests into one. Consequently, merging requests reduces overhead and minimizes seeks.

| b) The entire request queue is kept sorted, sectorwise, so that all seeking | activity along the queue moves (as much as possible) sequentially over the | sectors of the hard disk.This is similar to the algorithm employed in | elevators —— try to move gracefully in a single direction.

#. Linus Elevator: The Linus Elevator I/O scheduler performs both
   front and
   back merging.
#. The Deadline I/O scheduler: ensure that write requests do not
   starve
   read requests.
#. The Anticipatory I/O scheduler aims to continue to provide
   excellent
   read latency, but also provide excellent global throughput.
#. The Complete Fair Queuing (CFQ) I/O scheduler is an I/O scheduler
   designed for specialized workloads, but that in practice actually
   provides
   good performance across multiple workloads.It is now the default
   I/O scheduler
   in Linux(2.6).
#. the Noop I/O Scheduler truly is a noop, merely maintaining the
   request queue in near-FIFO order, from which the block device
   driver can pluck
   requests.

更多文章
  • Golang实现平滑重启(优雅重启)
  • traefik 教程
  • Web开发系列(十):事务和锁
  • Nginx作为TCP/UDP的负载均衡
  • Web开发系列(九):消息队列,异步任务
  • Web开发简介系列
  • Web开发系列(十一):数据库扩展
  • Nginx 请求匹配规则
  • Web开发系列(六):关系型数据库,ORM
  • Web开发系列(七):缓存,CDN
  • Web开发系列(八):单点故障,负载均衡
  • Web开发系列(五):form, json, xml
  • Web开发系列(四):Flask, Tornado和WSGI
  • Web开发系列(三):什么是HTML,CSS,JS?
  • Web开发系列(二):HTTP协议