Architecture - Database Management Systems - Lecture Slides, Slides of Introduction to Database Management Systems

Download Architecture - Database Management Systems - Lecture Slides and more Slides Introduction to Database Management Systems in PDF only on Docsity!

Database Management Systems Design

Relational DBMS Architecture

Disk Space Management

Buffer Management

File and Access Methods

Relational Operators

Query Optimizer

Query Parser

Client API

Client

DB

Execution Engine

Concurrency and Recovery

Disk Page

• Disk page is simply an array of bytes

• We impose the logic of an array of records!

123 Bob NY $ 2178 Jil LA $ 8273 Ned FL $ 723 Al PR $

Disk Page Records

Reading a Disk Page should be one I/O

Disk arrangement option

• Suppose we need to create 10 GB of space to stored a

database. Each page is 4 KB is size.

• How to organize the disk to accomplish this.
• Cooked File
• User the file system provide by OS
• Create a file “mydb.dat”
• Write to this file N pages of size 4KB
• N must be enough to reach the size of 10GB
• Page are full of bytes with zeros.
• Have a hast table somewhere to store the information about this file “mydb.dat”.
• Now you can start writing pages with actual data.

Buffer Management

• All Data Pages must be in memory in order to be

accessed.

• Buffer Manager

• deals with asking Disk Space Manager for pages from

disk and store them into memory

• Sends Disk Space Manager pages to be written to disk.

• Memory is faster that Disk

• Keep as much data as possible in memory
• If not enough space is available, need a policy to

decide what pages to remove from memory.

Buffer Pool

• Frame
  • Data strucuture that can hold a data page and control flags
• Buffer pool
  • Array of frames of size N.
• In C #define POOL_SIZE 100 #define PAGE_SIZE 4096 typedef struct frame { int pin_count; bool dirty; char page[PAGE_SIZE]; } frame; frame buffer_pool[POOL_SIZE];

Operational mode

• All requested data pages must first be placed into the

buffer pool.

• pin_count is used to keep track of number of transactions

that are using the page

• 0 means no body is using it
• dirty is used a flag (dirty bit) to indicate that a page has

been modified since read from disk

• Need to flush it to disk if the page is to be evicted from pool
• Page is an array of bytes where the actual is located
• Need to interpret this bytes as the int, char, Date data types supported by SQL
• This is very complex and tricky!

Buffer replacement

• If we need to bring a page from disk, we need

to find a frame in the buffer to hold it

• Buffer pool keeps track on the number of

frames in use

• List of frames that are free (Linked list of free frame nums)

• If there is a free frame, we use it

• Remove from list of free frame
• Increment the pin_count
• Store the data page into the byte array (page field)Docsity.com

Some issues

• Need to make sure pin_count is 0

• Nobody is using the frame

• Need to write the data to disk in dirty bit is

true

• This latter approach is called Lazy update

• Write to disk only when you have to!!!
• Careful, if power fails, you are in trouble.
• DBMS need to periodically flush pages to disk
  • Force write

• If no page is found with pin_count equal to 0,Docsity.com

Buffer Replacement policies

• LRU – Least Recently Used

• Evicts the page is the least recently used page in the pool.
• Can be implemented by having a queue with the frame numbers.
• Head of the queue is the LRU
• Each time a page is used it must be removed from current queue position and put back at the end - This queue need a method erase() that can erase stuff from the middle of the queue

• LRU is the most widely used policy for bufferDocsity.com