DISTRIBUTED DATABASE SYSTEMS - Database Design and Development: An Essential Guide for IT Professionals - page 585

• The system catalog or data dictionary resides in one place for a centralized
database. However, for a distributed database, a few options are available for
placing the catalog: at a central site, at all sites fully or partially, and a hybrid
option.
• Data distribution methods: data fragmentation (horizontal and vertical parti-
tioning) and data replication.
• Computing architectural options: two-tier client/server, two-tier intelligent
server, and three-tier architecture with middleware.
• Design and implementation issues focus on fragmentation and allocation of
data to the various sites.
• The key motivation for the DDBMS is transparency or hiding complexities of
data distribution and making it transparent to users. Transparencies provided
at different levels: fragmentation, replication, location, network, naming, and
failure.
• A query in a distributed environment typically gets divided into subqueries and
executed at the sites where the pertinent data are stored. Numerous additional
factors must be considered for query optimization.
• Transaction execution and propagation of updates to various replicas are more
difficult. Lock management and concurrency control pose many challenges.
• The two-phase commit protocol (2PC) is the industry standard to enable dis-
tributed recovery.
REVIEW QUESTIONS
1. Briefly describe the basic concepts of a distributed database system.
2. List five advantages and five disadvantages of distributed databases.
3. Compare and contrast homogeneous and heterogeneous systems.
4. What are the options for storing the system catalog or data dictionary?
5. Name any three of the network configurations for a distributed database
system. Describe one of these.
6. What is data fragmentation? Describe it using an example of a CUSTOMER
relation.
7. Name three types of transparencies provided by a DDBMS. Describe any
one.
8. List the four common techniques for update propagation.
9. Describe the essential differences between centralized and distributed
schemes for using wait-for graphs.
10. What is the underlying principle of the two-phase commit protocol?
EXERCISES
1. Indicate whether true or false:
A. In a distributed database system, each site must have autonomy to manage
its data.