• Data replication is common in distributed databases. Which site to query? Which site to update first and then the rest (with cascades)?
  • If a site (or a link) fails then, on re-connecting the isolated sites, the relevant data commits done during the interruption must be updated.
  • Transaction scheduling over multiple sites is much harder than the centralised case.
Print Friendly
 
  • Lack of experience or complete knowledge.
  • Complexity.
  • Costs.
  • Control.
  • Security.
  • Upgrading a database to a distributed database is not self evident.
Print Friendly
 
  • Performance.
  • Reliability.
  • Upsizing.
  • Shareability.
  • Local autonomy.
Print Friendly
 
  • A distributed database is a collection of a number of logically related databases partitioned over a computer communications network, but whose data is available to all of its end users, irrespective of their physical location.
  • A distributed DBMS maintains distributed databases and attempts to achieve a high level of distribution transparancy.
Print Friendly
 

Strange combination.

  • Databases are a logical repository of structured data. With most implementations having a centralised framework.
  • High level networks promote sharing.
Print Friendly
 

Could a database be made available to a number of users without specifying a "single" computer? – Yes.

 

How?

  • Distributing the data.
  • Distributing the DBMS processing.
  • Distributing the database control etc.

What do we need?

  • More refined database techniques.
  • Better and new DBMS technology.
  • A networking infrastructure that connects all of the databases' end users.
Print Friendly
 

If there are two fragments F and S, allocation problem involves finding optimal distribution of F to S.

Optimality can be defined;

  • Minimal cost.
  • Performance.
Print Friendly
 

Major information requirement is related to applications. The major requirement for applications is access frequencies.

Algorithms to calculate vertical partitioning include:

  • Clustering algorithm.
  • Partitioning algorithm.
Print Friendly
 

To obtain optimal solutions in vertical fragmentation, heuristics is used.

Two types of heuristics approach exists:

  • Grouping: assigns each attribute to one fragment, and at each step joins some of the fragments until some criteria is satisfied.
  • Splitting: starts with a relation and decides on beneficial partitionings based on the access behaviour applications to the attributes.

Splitting fits more naturally in the top-down methodology and the optimal solution is more likely found in a full relation.

Print Friendly
 

Horizontal fragmentation partitions a relation along its tuples. There are two versions of horizontal partitioning:

  • Primary horizontal fragmentation.
  • Derived horizontal fragmentation.

 

Primary horizontal fragmentation: is defined by a selection operation on the owner relations of a database schema.

Derived horizontal fragmentation: is defined on a member relation of a link according to a selection operation specified by its owner.

Print Friendly