File Organization and Storage Structures - 1
File Organization and
Storage Structures
File Organization and Storage Structures - 2
File Organization and Storage Structures
o
Storage of data
– Primary Storage = Main Memory
• Fast• Volatile • Expensive
– Secondary Storage = Files in disks or tapes
• Non-VolatileSecondary Storage is preferred for storing
data
Basic Concepts
o Information are stored in data files
o Each file is a sequence of records
o Each record consists of one or more fields
B3 WL220658D Deputy
Ford SG14
B3 WL432514C Snr Asst
Beech SG37
B5 WK440211B Manager
White SL21
Bno NIN
Position Lname
Sno
Logical Record Vs Physical Record
o Logical record
– Eg. The record of a staff (SG37). – “A record”
o Physical record
– The unit of transfer between disk and primary storage.
– “A page”, “A block”
File Organization and Storage Structures - 5
Logical Record Vs Physical Record
2
File Organization and Storage Structures - 6
File Organization & Access Method
o File Organization means the physical arrangement of data in a file into records and pages on
secondary storage
– Eg. Ordered files, indexed sequential file etc.
o Access Method means the steps involved in storing and retrieving records from a file.
– Eg. Using an indexed access method to retrieve a record from an indexed sequntial file.
Heap Files
o Heap files are files of unordered records.
o Quick insertion (no particular ordering)
– When a new record is created, it is put in the last page of the file if there is sufficient space. Otherwise a new page is added to the file.
o Slow retrieval (only allow linear search)
– reading pages from the file until a required record is found.
o To delete a record, the record is marked as deleted. Space is reclaimed during periodical reoganization.
Ordered Files
o Ordered Files: Records are sorted on field(s) => Key
o Allow Binary Searching
Suppose one page stores one record.
To search for SG37, search the middle page (6/2 = 3) first. We find that SG37 does not exist in this
File Organization and Storage Structures - 9
Ordered Files
o Inserting a record
– If the appropriate page is full, may have to re-organize the whole file => Time consuming
– Solution: use a temporary unsorted file (transaction file). Merge to the sorted file periodically.
o Rarely used unless come with an index => Indexed Sequential File
o Both Heap Files and Ordered Files are also called Sequential Files.
File Organization and Storage Structures - 10
Direct Files
o Direct Files are also called Hash Files or Random Files
o No need to write records sequentially
o Use a hash function to calculate the number of the page (bucket) which a record should be located
o Eg., use the division-remainder calculation method that,
bucket_no = Record_key mod 3
Direct Files
o Problem: If a new record SG41 is created, which bucket to go?
o Collision Management
Open addressing, Unchained overflow, Chained overflow, Multiple hashing
Direct Files
Open Addressing
o Upon a collision, the system performs a linear search to find the first available slot.
o When last bucket has been searched, starts from the first bucket.
File Organization and Storage Structures - 13
Direct Files
Unchained Overflow
o An overflow area is maintained for collisions.
o SL41 will be inserted to:
Bucket 3
File Organization and Storage Structures - 14
Direct Files
Chained Overflow
o Each bucket has a synonym pointer
o Value of the synonym pointer:
Zero: no collision occurred
Non-zero: the overflow bucket used
Direct Files
Multiple Hashing
o Upon collision, apply a second hashing function to produce a new hash address in an overflow area.
Direct Files
Limitation (of Hashing)
Inappropriate for some retrievals:
– based on pattern matching
eg. Find all students with ID like 98xxxxxx. – Involving ranges of values
File Organization and Storage Structures - 17
Indexes
Index: A data structure that allows particular records in a file to be located more quickly
~ Index in a book
An index can be sparse or dense:
Sparse: record for only some of the search key values
(eg. Staff Ids: CS001, EE001, MA001). Applicable to ordered data files only.
Dense: record for every search key value. (eg. Staff Ids: CS001, CS002, .. CS089, EE001, EE002, ..)
File Organization and Storage Structures - 18
Indexes
TERMINOLOGY
Data file: a file containing the logical records
Index file: a file containing the index records
Indexing field: the field used to order the index records in the index file
Key: One or more fields which can uniquely identify a record (eg. No 2 students have the same student ID).
Indexes
TYPES OF INDEXES
Primary Index: An index ordered in the same way as the data file, which is sequentially ordered according to a key. (The indexing field is equal to this key.)
Secondary Index: An index that is defined on a non-ordering field of the data file. (The indexing field need not contain unique values).
A data file can associate with at most one primary index plus several secondary indexes.
Indexed Sequential Files
What are Indexed Sequential Files?
= A sorted data file with a primary index
Advantage of an Indexed Sequential File
Allows both sequential processing and individual record retrieval through the index.
Structure of an Indexed Sequential File
o A primary storage area
o A separate index or indexes
File Organization and Storage Structures - 21
B
+-Trees
Point to
data
In B+-Tree, data or indexes are stored in a hierarchy of
nodes
File Organization and Storage Structures - 22
B
+-Trees
o B => Balanced
o Consistent access time (for each access, same number of nodes are searched)
TERMINOLOGY
Degree (Order) : The maximum number of children allowed per parent.
Depth : The maximum number of levels between the root node and a leaf node in the tree.
B
+-Trees
In practice, each node in the tree is actually a page, so we can store many pointers and keys. Eg. For a page size of 4KB, the B+-Tree can be of order 512.
Access time depends more ofen upon depth than on breadth => Shallow trees are preferred.
RULES
o The root (if not a leaf node) must have at least 2 children
o For a tree of order n, each node (except root and leaf) must have between n/2 and n pointers and children. If n/2 is not an integer, the result is rounded up.
B
+-Trees
RULES (Cont’d):
o For a tree or order n, the number of key values in a leaf node must be between (n-1)/2 and (n-1) pointers and children. If (n-1)/2 is not an integer, the result is rounded up.
o The number of key values contained in a nonleaf node is 1 less than the number of pointers.
o The tree must always be balanced: every path from the root node to a leaf must have the same length.
File Organization and Storage Structures - 25
B
+-Trees
Balancing can be costly to maintain.
Example:
Adding
SG14
File Organization and Storage Structures - 26