12/16/2014 1

I’m:

Iftikhar Ahmad Roll No: 13

From:

M-C-S (3rd sem)

AWKUM (shankar)

2

Presentation Topic

Distributed Database & Issues + Challenges

Homogeneous Database Heterogeneous Database

12/16/2014 3

Introduction:

A “Distributed Database”: is a logically interrelated collection of shared data (and a description of this data), physically distributed over a computer network.

A “Distributed DBMS” (DDBMS): is a Software system that permits the management of the distributed database and makes the distribution transparent to users.

Distributed Database

12/16/2014 4

DDBMS has following characteristics: o Collection of logically-related shared data. o Data split into fragments. o Sites linked by a communication network. o Data at each site is under control of a DBMS. o Each DBMS participates in at least one global application. o Data sharing o Data communication reliability and costs o Database recovery o Transaction and analytic processing

12/16/2014 5

Characteristics DDB

12/16/2014 6

Important difference between DDBMS and distributed processing !

DDBMS Distributed processing of centralised DBMS

12/16/2014 7

Why use a DDBMS? (!)

Advantages:

• Reflects organizational structure.

• Improved share ability.

• Improved availability.

• Improved reliability.

• Improved performance.

• Economics.

• Modular growth.

Disadvantages:

• Complexity.

• Cost.

• Security.

• Integrity control more difficult.

• Lack of standards.

• Lack of experience.

• Database design more complex.

12/16/2014 8

Types of Distributed Database System

DDBMS

Homogenous Heterogeneous

Homogeneous Heterogeneous

The prefix “homo” – indicate sameness

All sites use same DBMS product.

•

In this type of database has all data center have same software.

Much easier to design and manage. It appears to user as a single system. Allows increased performance. Much easier to design and manage.

12/16/2014 9

Homogeneous DB

12/16/2014 10 Same software

Homogeneous DB

Advantages of Homogeneous DB:

Easy to use Easy to mange Easy to Design

Disadvantages of Homogeneous DB: Difficult for most organizations to force a homogeneous environment.

12/16/2014 11

The prefix “hetero” – indicate difference

Sites may run different DBMS products, underlying data models.

In this type of database , Different data center may run different DBMS products.

Occurs when sites have implemented their own databases and integration is considered later. • Translations required to allow for:

• Different hardware. • Different DBMS products. • Different hardware and DBMS products.

12/16/2014 12

Heterogeneous DB

12/16/2014 13

Sql oracle

Heterogeneous DB

Advantages of Heterogeneous DB:

Huge data can be stored in one Global center from different data center.

Remote access is done using the global schema.

Different DBMSs may be used at each node.

Disadvantages of Heterogeneous DB:

Difficult to mange.

Difficult to design.

12/16/2014 14

Problems & their solutions:

The challenges encountered in distributed database transaction and proffered likely solutions. Few of the mentioned challenges are; problems of distribution of resources, search and updating of resources.

12/16/2014 15

Problems of Distributed Database (Transactions)

Distribution of resources problem: This has to do with equal distribution of resources across all servers. There are two approaches to solving this problem in a distributed database environment; decentralize by function, or decentralize by location.

(a) Decentralize by function: This works well with data that will be accessed repeatedly by the same users. Example includes putting manufacturing materials list at the appropriate manufacturing plants and customer information at sales locations.

(b) Decentralize by location: Occurs when you partition customer information on a node per region or other geographical based entity within an organisation as seen in figure.

12/16/2014 16

12/16/2014 17

Each telephone company independently implements the common protocols by the international phone network, e.g. for dialling and billing as shown in figure 2, the dialling and billing of customer vary from one country to another. The only centralize function is the architecture of these protocols. It is worthy to note that design and architecture are typically centralize within each company; while operation and control are delegated to the operating countries which in turn delegate operation and maintenance to individuals responsible for maintaining their own hardware and software.

12/16/2014 18

SEARCHING: Another notable problem of distributed database results from lack of adequate knowledge of the entire database. For example, locating where file x, y, z was stored. The solution to the challenge is using the concepts of global & local data. i. Global data involves information that is common to all sites and shared by all sites. ii. Local data is information that is strictly meant for an individual site although it is accessible to all sites. Information is made available to the whole network by partitioning or replicating the data files. Partitioning involves splitting a data file into records and then distributing the records across networks while replication means duplicating records at more than one location/node. Partitioning only works with local data while global data are replicated. Example i. A typical example is found in bank transactions where servers are located in local vicinity where customers can transact which often increases the speed of transaction when traffics are eradicated.

12/16/2014 19

Advantages of partitioning a network:

i. It makes data available to all remote users

ii. It reduces traffics and increase speed of access

Advantages of replicating a network:

i. Replication improves data availability

ii. Replication improves response time.

12/16/2014 20

UPDATING: As discussed above, partitioning of data is most efficient when the data are kept current that is, updated regularly. The single copy of each data item makes updating an efficient process. However. nonlocal “read” operations are more expensive, making partitioning less efficient for data that is widely used but updated infrequently. Replicated data is most efficient when multiple reads of the data are expected, but updates are not as regular compared to partitioning process. The data is duplicated at nodes where high- volume reads are expected, producing high availability and good response time. When replicated data must be updated, however, an update to a record at one node should cause an identical update at all other nodes where that record resides. If any one replica is unavailable there could be problems.

12/16/2014 21

A variety of schemes can be employed for updating Replicated data, even though the copy of the record may be temporarily unavailable at one or more of the nodes. One technique requires that a majority of the replicas be read and updated as part of each transaction, though the definition of “majority” varies with the application. This scheme has the advantage of tolerating some nodal unavailability, but it is not practical for either very small or very large networks. In a very small network of, say, two nodes, having either node unavailable prevents an update of a majority of the nodes. In larger networks, delays in completing the update transaction are proportional to network size: As the network grows, transactions will take longer to complete.

12/16/2014 22

Future work: The studies on researching new update strategies include the experimental distributed database management system extension by the implementation of the chosen update methods in the EDDBMS, work out of the replicated data update method in the VSM environment and its implementation. The next step will concentrate on setting the assumptions on the use of this algorithm in the other distributed programming environments and on consideration of the possibility of applying the system for building a data warehouse.

12/16/2014 23

The presentation has treated the proposed challenges of distributed database system & basics. However, new challenges show up daily in transactions. Therefore there may be need for constant reviewers of these challenges.

12/16/2014 24

CONCLUSION

Iftkr

iffi910

12/16/2014 25