Scribd
Upload
58 views3 pages

Optimizing Disk Data Organization

The document discusses reducing I/O costs by organizing data sequentially on disks. It notes that while disk capacity has increased significantly over time, transfer rates have not improved as much. This makes disk accesses more valuable. Software solutions to reduce I/O costs include arranging file blocks sequentially, reading/writing in larger chunks, and buffering. Hardware solutions involve distributing data across multiple disks to allow parallel access. The document provides an example comparing reading a file stored in large versus small pages to illustrate the impact of data organization.

Uploaded by

Mahesh Kumar Sah
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
58 views3 pages

Optimizing Disk Data Organization

The document discusses reducing I/O costs by organizing data sequentially on disks. It notes that while disk capacity has increased significantly over time, transfer rates have not improved as much. This makes disk accesses more valuable. Software solutions to reduce I/O costs include arranging file blocks sequentially, reading/writing in larger chunks, and buffering. Hardware solutions involve distributing data across multiple disks to allow parallel access. The document provides an example comparing reading a file stored in large versus small pages to illustrate the impact of data organization.

Uploaded by

Mahesh Kumar Sah
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Reducing I/O costs

How crucial is the problem?


Disk capacity has improved 1,000X in the last 15 year. The size of data also has increased in the same rate. But,
platters only spin 4X faster. the transfer rate has improved only 40X in the same period.

Data Organization on Disks


Davood Rafiei

Thus:
disk accesses are more precious. they are expected to be more precious in future.

Physical Disk Structure

Reducing I/O costs (Cont.)


Key to lower I/O cost: reduce seek/rotation delays Software solutions
arrange blocks of a file sequentially on disk read/write in bigger chunks buffering

Hardware solutions
Will not be discussed.

Reducing I/O costs (Cont.)


Store pages containing related information close together on disk
Justification: If application accesses x, it will next access data related to x with high probability

Buffering
Keep cache of recently accessed pages in main memory
Goal: request for page can be satisfied from cache instead of disk Purge pages when cache is full
For example, use LRU algorithm Record clean/dirty state of page (clean pages dont have to be written)

Page size tradeoff:


Large page size - data related to x stored in same page; hence additional page transfer can be avoided Small page size - reduce transfer time, reduce buffer size in main memory Typical page size - 4096 bytes

Example Page Size


Consider:
An IBM Deskstar disk with 40 sectors/track, 512 bytes/sector and average seek time of 9.1 msec. Disk platters spin at 7,200 rpm. average rotational delay=(1/7200)/2 minutes = 4.17 msec. transfer time for a sector= (1/7200)/40 minutes = 0.21 msec A file of 6400 256-byte records = 1638 KB which occupies 3200 sectors of the disk.

Accessing Data Through Cache


DBMS Page transfer

Application

cache/buffer

Case 1:
The file is stored in 100 extents each of size 4 pages where each page is 8 sectors. Time to read the file= 100 x ( 9.1 + 4.17 + 32 x 0.21)=2 seconds

Case 2:
The file is stored in 3200 pages each of size one sector. Time to read the file = 3200 (9.1 + 4.17 + 0.21) = 43 seconds
6

Item transfer

block Page frames

Hardware Solutions
Arrange disks arrays: several disks that give abstraction of a single, large disk. Partition data into striping units and distribute them over several disks.
student file:
pages 1-10 pages 11-20 pages 21-30

read pages 1,22,15

more disks --> more failures!


9

Summary
Disks: cheap, non-volatile storage.
provides both sequential and random access. The cost for a random access depends on the location of page on disk; important to arrange data sequentially to minimize seek and rotation delays.

Lowering I/O costs


software vs. hardware solutions.

10

You might also like