1. How can replication help the performance of a distributed database?
2. In what situations can replication hurt the performance of a distributed database?
The following four exercises are related. The first three are about fragmentation, allocation, and replication considerations in DDBMS design, and the fourth asks you how these interact in the design process.
3. What are the main considerations that drive the fragmentation design of a distributed database? (short answer with several parts)
4. What are the main considerations that drive the allocation design of a distributed database? (short answer with several parts)
5. What are the main considerations that drive the replication design of a distributed database? (short answer with several parts)
6. How do fragmentation, allocation and replication interact in the design of distributed databases? (short answer with several parts)
For the following four exercises: A SALES table contains financial, sales office, and product data for a company that has computer facilities and databases at their corporate headquarters and regional sales offices. Billing, accounting, financial analysis and other financial transactions are performed at the headquarters computer facility and sales functions are performed at the regional office computer facilities. Evaluate the following fragmentation strategies for distributing the SALES table, listing advantages and disadvantages.
7. Horizontal fragmentation to the regional offices, with no replication. Financial data is transmitted each night from the regional offices to headquarters.
Advantages:
Disadvantages:
8. Horizontal fragmentation by geographic region, with full replication of the regional data at headquarters. The replication is configured so that the regional offices are fully functional without the connection to headquarters, and the replica at headquarters is synchronized when connections to the field are restored.
Advantages:
Disadvantages:
1. Slow down update operations
9. Horizontal fragmentation by geographic region, with replication of the financial and other critical data at headquarters. The replication is configured so that the regional offices are fully functional without the connection to headquarters, and the replica at headquarters is synchronized when connections to the field are restored.
Advantages:
Disadvantages:
10. Vertical fragmentation, with financial columns at headquarters, and all other columns at regional sales offices. The regional and headquarters columns are joined by a view that is used by the applications.
Advantages:
Disadvantages:
11. Which of the above designs do you think is best for a very large global enterprise, and why? (short answer, potentially not conclusive)
12. Provide a database distribution design which you may think is better for the above application and explain why.