Advanced Database Lecture Notes

Date: 11/1/2013 Advanced Database Design Lectures Note Jasour Obeidat
Chapter 17: Physical DB Design for Relational Q7. What are the steps involved in the
DB Transparencies translation of logical data model to target
DBMS?
Q1: What are the sources of physical design? ANS:
ANS: 1. Design base relation.
1. Logical Data Model. 2. Design a representation of the derived
2. Documentation that describe the model. data.
Q2: Multiple Choice Questions (MCQ) 3. Design general constraints.
- DB design that concerned with 'WHAT' is: Q8. Why we need to design a base relation
a. Logical b. Physical c. Conceptual step?
ANS: (a) ANS:
- Physical DB design concerned with: To decide how to represent the base relations
a. WHAT b. HOW c. Who identified in the logical data model in the target
ANS: (b) DBMS.
Q3. Define the term of Physical Design? Q9. In design base relation step, for each
ANS: relation what we have to define?
Is the process of producing a description of the ANS:
implementation of the database in the secondary 1. The name of relation.
storage 2. The list of simple attributes in brackets.
Q4. What did Physical design describe? 3. The primary key, auxiliary key, and FK.
1. Base Relations. 4. Referential integrity for each FK
2. File organization. identified in the relation.
3. Indexes used to achieve efficient access to Q10. In design base relation, for each attribute
data. from data dictionary, what we have to define?
4. Integrity constraints. ANS:
5. Security measures. 1. Attribute domains { data types, length ,
Q5. Describe the goal from translating logical domain constraints}.
data model for target DBMS? 2. Optional And/ Or default values of
ANS: attribute in the relation, and whether it can
To produce a relational database schema from the hold NULL.
logical data model that can be implemented in the 3. Whether attribute is derived and if so how
target DBMS it is computed?
Q6. Why we need to know the functionality of Q11. Why we need to design a representation
target DBMS? of derived data?
ANS: ANS:
1. To know how to create a base relations. To decide how to represent the derived data
2. To know whether it support a definition identified in the logical data model in the target
for primary, auxiliary, secondary, and DBMS.
foreign keys. Q12. How to design a representation for
3. To know whether it is support domains. derived data?
4. To know whether it is support general ANS:
constraints. 1. By examining the logical data model and
5. To know whether it is support integrity data dictionary we can produce a list of
constraints. derived attributes.
6. To know whether is support NOT, NULL

Middle East University of Jordan (MEU) jasour_2004@yahoo.com

2. Derived attributes have to options, either - By using peak load which it is the
storing them into database or calculate during time of attribute / relation of
them every time and it is required. when will be the high demand of
3. The chosen option based on: database.
- If we store the derived data in the 2. By using transaction analysis information
relation and keep it consistent with the to identify the parts of database that cause
operational data which it derived performance problems.
from. 3. Need to identify the high level of
- The cost of calculated the derived data functionality of transaction such as
every time. attribute update, search criteria used in
4. Less expensive option based on: chosen a query.
subject based on performance constraints. 4. Often we will not analyze all transactions,
Q13. Why we need to design general so by investigating the important ones by:
constraints? - Using the attribute/ relation cross
ANS: reference matrix; which show the
Because some DBMS provide facilities than relation accessed by each transaction.
others in defining enterprise general constraints - Using the usage map; which show the
heavily used relations.
PART TWO 5. Focus on the parts of database may be
Q1. Why we need to define file organization problematic by:
and indexes? - Map the path between transaction to
ANS: relations.
1. To determine the optimal file organization - Determine the relations that are
to store the base relations. frequently accessed by transactions.
2. To determine the indexes to achieve an - Analyze the data usage of selected
acceptable performance; 'The way that transaction that involves these
tuples and relations stored in secondary relations.
storage' Q4. Why we need to identify file organization?
Q2. What are the steps involved in design file ANS:
organization and indexes? We need to choose specific file organization in
ANS: order to:
1. Analyze transaction. - Determine the efficient file
2. Choose file organization. organization for each base relation
3. Choose indexes. such as using:
4. Estimate disk space requirements. i. Heap
Q3. How to analyze transactions? ii. Hash
ANS: iii. Indexed Sequential Access
1. Attempt to identify performance criteria Method (ISAM).
such as: iv. Clusters.
- Transactions that runs frequently on v. B+ Trees.
relations and have a significant impact Q5. TRUE or FALSE Question.
on performance. ( ) Most of DBMS may not allow choosing or
- Transactions that is critical to selecting file organization.
organization. ANS: TRUE.


PART III Q7. By balancing the overhead of maintenance
Q1. Why we need to choose indexes? for secondary index against performance
ANS: improvement gained in retrieving data, what
We need it to determine whether adding indexes this means? Or what this includes?
to relation will improve the performance over the ANS:
database. 1. Adding secondary record to every
Q2. Mention the both approaches used in secondary index when a new tuple
choosing indexes? inserted.
ANS: 2. Increase the disk space to store secondary
1. One approach is to keep tuples in the index.
relation unordered and add secondary 3. Update secondary index whenever the
indexes as necessary. corresponding tuple updated.
2. Another approach is to order tuples in the 4. Possible for performance degradation
relation by specifying primary index or while making query optimization in order
clustering index. to consider all possible secondary indexes.
Q3. In the approach of specifying primary or Q8. Mention the guidelines for making 'wish-
clustering index, what we have to do? list'?
ANS: ANS:
By choosing attribute for ordering or clustering 1. Avoid index small relations.
based on: 2. Index primary key if it is not used as a key
- Attribute often used for JOIN in file organization.
operation in order to make JOIN 3. Add secondary index for attributes used in
operation more efficient. built in functions.
- Attribute used for accessing relation 4. Add secondary index for attributes
tuples in order of that attribute. involved in SELECT, JOIN, ORDERED
Q4. MCQ BY, GROUB BY operations.
- If attribute chosen was primary for the 5. Add secondary index for foreign key
relation the index called: accessed frequently in the relations.
a. Primary Index b. Clustering Index 6. Avoid index attribute updated frequently.
ANS: (a) 7. Avoid index attribute contain domain with
- If attribute chosen was not primary key a long string 'characters'.
attribute, the index in this case is: 8. Add secondary index for attribute used in
a. Primary Index b. Clustering Index. index-only plan
ANS: (b) 9. Avoid index for attribute involved in
query to retrieve a significant proportion
Q5. TRUE or FALSE Question of the relation.
( ) – Each relation can have either primary 10. Add secondary index for attribute heavily
index or Clustering index. used as a secondary key in the relation.
ANS: TRUE

Q6. What does Secondary Index provide?
ANS:
Provide a mechanism for specifying an additional
key for base relation to make retrieving data more
efficient.


Chapter 22: Distributed Database & DDBMS Q8. Mention the advantages of DDBMS?
ANS:
Q1. What is distributed database? 1. Economic
ANS: 2. Reflect the organizational structure.
A logically interrelated shared date (and the 3. Improved performance.
description of this data) physically distributed 4. Improved availability.
over a computer network. 5. Improved reliability.
6. Improved shared-ability and local
Q2. What is distributed DBMS? autonomy.
ANS: 7. Modular Growth.
Software that permits the management of
distributed database and make the distribution Q9. Mention the disadvantages of DDBMS?
transparent to users. ANS:
1. Cost.
Q3. What distributed Processing? 2. Security.
ANS: 3. Complexity.
A centralized database can be accessed over a 4. Lack of standards.
computer network. 5. Lack of experience.
6. Database design more complex.
Q4. What is Parallel DBMS? 7. Integrity control is difficult.
ANS:
A DBMS running across multiple processors and Q10. Mention the types of DDBMS?
disks designed to execute operations in parallel, ANS:
whenever possible, to improve performance. 1. Homogenous DDBMS.
2. Heterogeneous DDBMS.
Q5. Why we need to Parallel DBMS?
ANS: Q11. Define formally Homogenous DDBMS?
Based on single processor it will not meet the ANS:
requirements of - All sites have the same DBMS
- Reliability. product.
- Scalability. - This approach support incremental
- Cost effective. growth, and increases performance.
- Performance. - Much easier to design and manage.
Q6. What is the idea behind Parallel DBMS? Q12. Define formally Heterogeneous DDBMS?
ANS: ANS:
Parallel DBMS link multiple, smaller machines to - Each site have different DBMS
meet the same throughput of single, larger product, possibly different data model.
machine with greater scalability and reliability. - Occurs when each site have already
implemented their own database, and
Q7. Mention the architectures used in Parallel integration considered later.
DBMS? - Translation should allow to have:
ANS: 1. Different H/W.
- Shared Memory. 2. Different DBMS product.
- Shared Disk. 3. Different H/W and DBMS product
- Shared Nothing. - Typically solution by using gateways.


PART II: Distributed DB design.

Q1. What are the key issues in DDB design? Q5. Define formally 'Completeness' rule?
ANS: ANS:
1. Fragmentation: a relation may be divided into If relation R decomposed into R1,R2,...,Rn each
sub-relations and distributed over sites. tuple in R should be found in at least one
2. Allocation: Each fragment is stored over sites fragment.
with 'optimal' distribution.
3. Replication: A copy of fragment that may be Q6. Define formally 'Reconstruction' rule?
maintained at several sites. ANS:
It is possible to define a relational operation that
Q2. Why we need to fragment? able to reconstruct relation R where:
ANS: - In VF the operation is JOIN
1. Usage: - In HF the operation is UNION
- Working over fragments such as working with
view rather than the entire relation. Q7. Define formally 'Disjointness' rule?
2. Efficiency: ANS:
- Data is stored close to where it is frequently If a data item di found in a fragment Ri it should
used. not appear in another fragment except in case of
- Data is not needed by the local application is not vertical fragmentation which will repeat the
stored. attribute of primary key.
3. Parallelism: Noting That:
- By working with fragments which represent a - In Vertical fragmentation data item is
unit of distribution, transaction could be divided attribute.
to sub queries and implemented over fragments. - In Horizontal fragmentation data item
4. Security: is tuple.
- Data is not needed by the local application is not Q8. Mention the types of fragmentation?
stored and is not available to unauthorized users. ANS:
1. Vertical fragmentation.
Q3. Mention the disadvantages of fragment? 2. Horizontal fragmentation.
ANS: 3. Mixed fragmentation.
1. Performance. 4. Derived fragmentation.
2. Integrity. Q9. TRUE or FALSE question
( ) if the relation is small it is recommended
Q4. What are the correctness rules in not to fragment this relation.
fragmentation? ANS: TRUE
ANS:
1. Completeness.
2. Reconstruction.
3. Disjointness.


Chapter 14: Indexing Structure for files 4. Include index entry of key field value of
PART I: ELMASRI Edition Contents the first record in block which called
________________________________________ 'block anchor'.
Q1. What are the types of single level index 5. It is an example of non-dense (sparse)
used? index because there are index entry for
ANS: each block in data file and the key of this
- Primary Index. block which it block anchor.
- Secondary Index.
- Clustering Index. Q8. Mention the characteristics of Clustering
Q2. Define the term 'Single Level Index'? Index?
ANS: ANS:
Is auxiliary files that make accessing data file and 1. Defined on an ordered data file.
searching for certain record in the file more 2. Data file ordered on a non key field. And
efficient this requires that this non key field
Q3. Define the basics of Index and its form? contains a distinct value for each record in
ANS: data file.
1. Index may be for one field in data file 3. Include one index entry for each distinct
2. Index may be for several fields in the data file. value of field.
4. Index entry points on the first data block
* The General Syntax for Index as the following: that contains that distinct value.
<Field Value, Pointer to Record> 5. It is an example of a non dense index.
* Index = Access Path on the field.
Q4. Why index file occupies less disk block Q9. Mention the characteristics of Secondary
than data file? Index?
ANS: ANS:
Because index entries is much smaller 1. Secondary index provide a secondary
Q5. What is yield from binary search over mean of accessing a data file which have a
index file? primary access exist.
ANS: 2. Secondary index may be for a candidate
Yields pointers to file records key that contain a unique value for each
Q6. Mention the characteristics of indexing? record in the data file or for a non key
ANS: field that contain a duplicate values in
- Dense Index: There is an index entry data file.
for each search value in data file. 3. Index file contain two fields:
- Non-Dense (Sparse) Index: there is an a. The first field contains the same data
index entries for some of search type of unordered field in data file.
values in data file. b. The second field contains either a
Q7. Mention the characteristics of Primary record pointer or block pointer.
Index? 4. Include index entry for each record in the
ANS: data file so it an example of dense index.
1. Defined on an ordered file.
2. Data file ordered on key field.
3. Include index entry for each block in data
file.


PART II: Multi Level Index

Q1. What is the idea behind the multi-level
index?
1. Because of single level index based on
ordered files, we can create an index for
the index itself. So we can call the original
index as first level, and the index of index
the second level of index.
2. We can repeat this process to have
second, third….etc level of index until all
index entries fit one disk block.
3. Multi level index could be used for any
type of index such as primary, clustering,
and secondary index while the first level
consist of more than one block.

Q2. MCQ

Multi Level Index is a form of:
a. Search Tree b. B-Tree c. B+-tree
ANS: (a)

Q3. True OR False Question, Why???

( ) – Insertion or deletion of new index entries
may not cause a problem in Multi-level index.
ANS: (False)
Reason IF False:
It causes a problem because every level is an
ordered file.

Q4. What is the difference between (B-Tree)
and (B+-Tree)?
ANS:
1. In B-Tree: Pointers to data records exist at
all levels of the tree.
2. In B+-Tree: Pointers to data records at the
leaf nodes only.
3. B+-Tree can have less level than the
corresponding B-Tree.
4. B+-Tree can have higher capacity of
search records than in B-Tree.


Advanced Database Lecture Notes

Recomendados

Recomendados

Más contenido relacionado

La actualidad más candente

La actualidad más candente (20)

Destacado

Destacado (20)

Similar a Advanced Database Lecture Notes

Similar a Advanced Database Lecture Notes (20)

Último

Último (20)

Advanced Database Lecture Notes