Chapter 6 Physical Database Design Physical Design u

Chapter 6 Physical Database Design

Physical Design u The purpose of the physical design process is to translate the logical description of the data into technical specifications for storing and retrieving data u Goal: create a design that will provide adequate performance and insure database integrity, security, and recoverability u Decisions made in this phase have a major impact on data accessibility, response times, security, and user friendliness.

Required Inputs u Normalized relations u Data volume and use estimates u Attribute definitions u Descriptions of where and when data are used u Expectations for response time, data security, backup, recovery, retention, and integrity u Description of chosen technology

Determining volume and usage u Data volume statistics represent the size of the business • calculated assuming business growth over a period of several years u Usage is estimated from the timing of events, transaction volumes, and reporting and query activity. • Less precise than volume statistics

Composite usage map (Pine Valley Furniture Company)

Physical Design Decisions u Specify the data type for each attribute from the logical data model • minimize storage space and maximize integrity u Specify physical records by grouping attributes from the logical data model u Specify the file organization technique to use for physical storage of data records u Specify indexes to optimize data retrieval u Specify query optimization strategies

Data Format Data type selection goals • minimize storage • represent all possible values – eliminate illegal values • improve integrity • support manipulation Note: these have different relative importance

Data format decisions (coding) u e. g. , AH(Adams Hall), B(Buchanan) , etc u implement by creating a look-up table u there is a trade-off in that you must create and store a second table and you must access this table to look up the code value u consider using when a field has a limited number of possible values, each of which occupies a relatively large amount of space, and the number of records is large and/or the number of record accesses is small

Example code-look-up table (Pine Valley Furniture Company)

Data Format decisions (integrity) Data integrity controls • default value • Range control • Null value control • Referential integrity Missing data • substitute an estimate • report missing data • Sensitivity testing

For example. . . u Suppose you were designing the age field in a student record at your university. What decisions would you make about: • data type • integrity (range, default, null) • How might your decision vary by other characteristics about the student such as degree sought?

Attribute groupings u Physical Record: A group of fields stored in adjacent memory locations and retrieved together as a unit. u Page: The amount of data read or written in one I/O operation. u Blocking Factor: The number of physical records per page.

Database Access Model The goal in structuring physical records is to minimize performance bottlenecks resulting from disk accesses (accessing data from disk is slow compared to main memory)

Attribute grouping: Denormalization u Process of transforming normalized relations into denormalized physical record specifications • may partition a relation into more than one physical record • may combine attributes from different relations into one physical record

Denormalization Involves a trade-off: 4 Reduced disk accesses and greater performance (due, for example, to fewer table joins) - But - 8 Introduction of anomalies (and thus redundancies) that will necessitate extra data maintenance 8 increase chance of errors and force reprogramming when business rules change 8 may optimize certain tasks at the expense of others (if activities change, benefits may no longer exist)

Denormalization opportunities u 1: 1 relationship u M: M associative entity with non-key attributes u reference data

A possible denormalization situation: reference data

More denormalization options u Horizontal Partitioning: Distributing the rows of a table into several separate files. u Vertical Partitioning: Distributing the columns of a table into several separate files. • The primary key must be repeated in each file. u Combination of both

Partitioning u Advantages of Partitioning: • Records used together are grouped together • Each partition can be optimized for performance • Security and recovery • Partitions stored on different disks: less contention • Parallel processing capability u Disadvantages of Partitioning: • Slower retrievals when across partitions • Complexity for application programmers • Anomalies and extra storage space requirements due to duplication of data across partitions

How about this. . Consider the following normalized relations: STORE(Store_Id, Region, Manager_Id, Square_Feet) EMPLOYEE(Emp_Id, Store_Id, Name, Address) DEPARTMENT(Dept#, Store_ID, Manager_Id, Sales_Goal) SCHEDULE(Dept#, Emp_Id, Date, hours) What opportunities might exist for denormalization?

Physical Files u Physical File: A file as stored on disk u Constructs to link two pieces of data: • Sequential storage • Pointers u File Organization: How the files are arranged on the disk (more on this later) u Access Method: How the data can be retrieved based on the file organization • Relative - data accessed as an offset from the most recently referenced point in secondary memory • Direct - data accessed as a result of a calculation to generate the beginning address of a record

File Organizations u A technique for physically arranging the records of a file on secondary storage devices. u Goals in selecting: (trade-offs exist, of course) • Fast data retrieval • High throughput for input and maintenance • Efficient use of storage space • Protection from failures or data loss • Minimal need for reorganization • Accommodation for growth • Security from unauthorized use

File Organizations u Sequential u Indexed • Indexed Sequential • Indexed Nonsequential u Hashed u See (also called Direct) Table 6 -3 for comparison

Sequential File Organization u. Records of the file are stored in sequence by the primary key field values

Comparisons of file organizations: (a) Sequential

Sequential Retrieval u Consider a file of 10, 000 records each occupying 1 page u Queries that require processing all records will require 10, 000 accesses u e. g. , Find all items of type 'E' u Many disk accesses are wasted if few records meet the condition u However, very effective if most or all records will be accessed (e. g. , payroll)

Indexed File Organization u. Index concept is like index in a book u. Indexed-sequential file organization: The records are stored sequentially by primary key values and there is an index built on the primary key field (and possibly indexes built on other fields, also)

(b) Indexed

Hashed File Organization u. Hashing Algorithm: Converts a primary key value into a record address u. Division-remainder hashing algorithm (more to come on this) method is common

(c ) Hashed

Hashing u A technique for reducing disk accesses for direct access u Avoids an index u Number of accesses per record can be close to one u The hash field is converted to a hash address by a hash function

Hashing

Shortcomings of Hashing u Different hash fields may convert to the same hash address • these are called Synonyms • store the colliding record in an overflow area u Long synonym chains degrade performance u There can be only one hash field per record u The file can no longer be processed sequentially

Clustering u In some relational DBMS, related records from different tables that are often retrieved together can be stored close together on disk u Because the related records are stored close to one another on the physical disk, less time is needed to retrieve the data • E. g. , Customer data and Order data may frequently be retrieved together u Can require substantial maintenance if the clustered data changes frequently

Indexing u An index is a table file that is used to determine the location of rows in another file that satisfy some condition

Querying with an Index u Read the index into memory u Search the index to find records meeting the condition u Access only those records containing required data u Disk accesses are substantially reduced when the query involves few records

Maintaining an Index u Adding a record requires at least two disk accesses: • Update the file • Update the index u Trade-off: 4 Faster queries 8 Slower maintenance (additions, deletions, and updates of records) • Thus, more static databases benefit more overall

Rules of Thumb for Using Indexes 1. Indexes are most useful on larger tables 2. Index the primary key of each table (may be automatic, as in Access) 3. Indexes are useful on search fields (WHERE) 4. Indexes are also useful on fields used for sorting (ORDER BY) and categorizing (GROUP BY) 5. Most useful to index on a field when there are many different values for that field

Rules of Thumb for Using Indexes 6. Find out the limits placed on indexing by your DBMS (Access allows 32 indexes per table, and no index may contain more than 10 fields) 7. Depending on the DBMS, null values may not be referenced from an index (thus, rows with a null value in the field that is indexed may not be found by a search using the index)

Rules for Adding Derived Columns u Use when aggregate values are regularly retrieved. u Use when aggregate values are costly to calculate. u Permit updating only of source data. u Create triggers to cascade changes from source data.

One Other Rule of Thumb for Increasing Performance u. Consider contriving a shorter field or selecting another candidate key to substitute for a long, multi-field primary key (and all associated foreign keys)

Query Optimizer Factors u Type of Query • Highly selective. • All or most of the records of a file. u Unique fields u Size of files u Indexes u Join Method • Nested-Loop • Merge-Scan (Both files must be ordered or indexed on the join columns. )

More practice u Draw a composite usage map for the following: • PERSON(person_ID, name, address, DOB) • PATIENT(PA_person_ID, Contact) • PHYSICIAN(PH_person_ID, specialty) • PERFORMANCE(PA_person_ID, PH_person_ID, Treatment#, Treatment_date, Treatment_time) • CONSUMPTION(PA_person_ID, Item#, Date, Quantity) • ITEM(Item#, Description) u Make recommendations about denormalizing, partitioning, file organization, and indexing

RAID u Redundant Arrays of Inexpensive Disks u Exploits economies of scale of disk manufacturing for the computer market u Can give greater security u Increases fault tolerance of systems u Not a replacement for regular backup

RAID u The operating system sees a set of physical drives as one logical drive u Data are distributed across physical drives u All levels, except 0, have data redundancy or error-correction features u Parity codes or redundant data are used for data recovery

Mirroring u u Write • Identical copies of file are written to each drive in array Read • Alternate pages are read simultaneously from each drive • Pages put together in memory • Access time is reduced by approximately the number of disks in the array Read error • Read required page from another drive Tradeoffs 4 Provides data security 4 Reduces access time 8 Uses more disk space

Mirroring Complete Data Set No parity

Striping u u u Three drive model Write • Half of file to first drive • Half of file to second drive • Parity bit to third drive Read • Portions from each drive are put together in memory Read error • Lost bits are reconstructed from third drive’s parity data Tradeoffs 4 Provides data security 4 Uses less storage space than mirroring 8 Not as fast as mirroring

Striping One-Half Data Set Parity Codes

Database Architectures u Hierarchical u Network u Relational u Object-oriented u Multidimensional

Hierarchical models u Type of logical database model in which data is organized in a tree structure. u Records are divided into segments which are linked using pointers. u Parent and child data are stored together.

Sample hierarchical model ROOT/PARENT FIRST CHILD 2 nd CHILD Employee Compensation Ratings Salary Job Assignments Pension Benefits Insurance Health

Network model u Extends the hierarchical model by allowing a child to have zero, one, or many parents u No real theoretical base u Never used widely CLASS 1 STUDENT A CLASS 2 STUDENT B STUDENT C

Comparing the alternatives ** OODBS are marketed as easy to use, but the transition from RDBMS is difficult