Transcript Disks

Disk Scheduling
 Disk IO requests are for blocks, by number
 Block requests come in an ongoing stream from apps
 Requests are queued (possibly blocking the process) until
the disk is available
 OS can optimize disk performance by reordering the queue
 Require low access time and high disk bandwidth.
 Access time is the time to start transferring data
 Bandwidth is the average transfer rate, from request to
completion
 Access time has two major components
 Seek time is the time for the disk are to move the heads to
the cylinder containing the desired sector. (large)
 Rotational latency is the additional time waiting for the disk
to rotate the desired sector to the disk head. (small)
 Minimize seek time  seek distance
Operating System Concepts
13.1
Silberschatz, Galvin and Gagne 2002
Disk Scheduling (Cont.)
 Given a request queue of block numbers, minimize
summed seek distance
 Block numbers are translated into track and sector numbers
 Higher block numbers correspond to higher track numbers
 Sector numbers may not be known exactly, but can be
computed for analysis
 Example, for a disk with range 0-199:
98, 183, 37, 122, 14, 124, 65, 67
 Head pointer initially at 53
 Optimal service order is
37, 14, 65, 67, 98, 122, 124, 183 (208)
Operating System Concepts
13.2
Silberschatz, Galvin and Gagne 2002
FCFS
Illustration shows total head movement of 640 cylinders.
Operating System Concepts
13.3
Silberschatz, Galvin and Gagne 2002
SSTF
 Selects the request with the minimum seek time from the
current head position.
 Illustration shows total head movement of 236 cylinders.
 SSTF scheduling is a form of SJF scheduling; may cause
starvation of some requests.
 Not optimal
Operating System Concepts
13.4
Silberschatz, Galvin and Gagne 2002
SCAN
 The disk arm starts at one end of the disk, and moves
toward the other end, servicing requests until it gets to the
other end of the disk, where the head movement is
reversed and servicing continues.
 Sometimes called the elevator algorithm.
 Illustration shows total head movement of 236 cylinders.
 Waste at far ends
Operating System Concepts
13.5
Silberschatz, Galvin and Gagne 2002
LOOK
 Version of SCAN
 Arm only goes as far as the last request in each direction,
then reverses direction immediately, without first going all
the way to the end of the disk.
Operating System Concepts
13.6
Silberschatz, Galvin and Gagne 2002
C-SCAN
 The head moves from one end of the disk to the other.
servicing requests as it goes. When it reaches the other
end, however, it immediately returns to the beginning of
the disk, without servicing any requests on the return trip.
 Treats the cylinders as a circular list that wraps around
from the last cylinder to the first one.
 Illustration shows total head movement of 183 cylinders,
plus return time.
 Provides a more uniform wait time than SCAN.
Operating System Concepts
13.7
Silberschatz, Galvin and Gagne 2002
C-LOOK
 Version of C-SCAN
 Arm only goes as far as the last request in each direction,
then reverses direction immediately, without first going all
the way to the end of the disk.
 Illustration shows total head movement of 153 cylinders,
plus return time.
Operating System Concepts
13.8
Silberschatz, Galvin and Gagne 2002
Selecting a Disk-Scheduling Algorithm
 Performance depends on the number and types of requests.
 SSTF is common and has a natural appeal, but can starve requests
 SCAN and C-SCAN perform better for systems that place a heavy
load on the disk - lots of requests to hit
 Optimal is NP-complete.
 Requests for disk service can be influenced by the file-allocation
method and directory layout.
 The disk-scheduling algorithm should be written as a separate
module of the operating system, allowing it to be replaced with a
different algorithm if necessary.
 Either SSTF or LOOK is a reasonable choice for the default
algorithm.
 Newer disks have
 Lower seek time relative to latency
 Built in scheduling - send requests in batches
 OS level scheduling is important at a coarser grain level, for
important disk activity, e.g., VM
Operating System Concepts
13.9
Silberschatz, Galvin and Gagne 2002
Bad Blocks
 Disks come from the factory with bad blocks
 File system solutions
 Keep a bad blocks list, etc
 Low level solutions
 Sector sparing replaces bad blocks with spares, by a
mapping in the disk controller
 Sector slipping slides them down
 Interferes with disk scheduling - that’s why disks might be
left to their own devices (ha ha)
Operating System Concepts
13.10
Silberschatz, Galvin and Gagne 2002
Swap-Space Management
 Swap-space — Virtual memory uses disk space as an
extension of main memory.
 Swap-space can be
 Carved out of the normal file system
 In a separate disk partition with a special file system (no
directory, etc)
 Combination of these, e.g., Solaris, which always rereads
text pages from the file system (they’re never dirty, and
hence never swapped out)
 Swap-space management
 4.3BSD allocates swap space when process starts; holds
text segment and data segment.
 Solaris 2 allocates swap space only when a page is forced
out of physical memory, not when the virtual memory page
is first created.
Operating System Concepts
13.11
Silberschatz, Galvin and Gagne 2002
RAID
 RAID schemes improve performance and improve the
reliability of the storage system by storing redundant data.
 Disk striping uses a group of disks as one storage unit
 Bit striping gives parallel transfer for all size transfer
 Block striping gives parallel transfer for large transfers,
and concurrency for small accesses
 Mirroring or shadowing keeps duplicate of each disk.
 Block interleaved parity combines the ideas
 RAID is arranged into six different levels.
Operating System Concepts
13.12
Silberschatz, Galvin and Gagne 2002
RAID Levels
Operating System Concepts
13.13
Silberschatz, Galvin and Gagne 2002
RAID (0 + 1) and (1 + 0)
Operating System Concepts
13.14
Silberschatz, Galvin and Gagne 2002