Distributed Databases Questions and Answers

Answer: D

Correct answer: All of the above

All three are legitimate distribution strategies, chosen according to where the data is used. Centralised suits a system with one authoritative source; replication suits read-mostly data needed everywhere; and partitioning suits data naturally divided by site, such as regional customers. Most real systems combine them.

Answer: A

Correct answer: Data Replication

Replication stores a full copy at each of several sites, so every query can be answered locally and the failure of one site does not stop the others. The cost is keeping the copies consistent, which is why it suits data that is largely read and rarely changed.

Answer: C

Correct answer: Vertical Partitioning

Vertical partitioning divides a relation by column, keeping the primary key with each fragment so the original can be reconstructed by a join. It suits a case where different sites use different attributes of the same entity - personnel data at one site and payroll data at another, for the same employees.

Answer: A

Correct answer: The same DBMS is at each node and each DBMS works independently.

Homogeneous means the same product at every site; autonomous means each site manages itself, with no central authority coordinating access. The combination is common where an organisation has standardised on one product but its sites were developed separately and retain local control.

Answer: A

Correct answer: True

The semijoin sends only the joining column values out and receives only the matching rows back, so both directions carry far less than a full transfer of either table. Where the network is the bottleneck - which in a distributed database it usually is - that saving governs the cost of the query.

Answer: A

Correct answer: True

Synchronous replication updates every copy within the same transaction, which guarantees consistency but makes the transaction depend on every site being available. Asynchronous replication propagates changes afterwards, accepting a period of inconsistency in exchange for availability and speed, and that trade is the one usually taken.

Answer: C

Correct answer: The application's data can be somewhat out-of-date.

Replication is asynchronous in practice, so copies lag behind one another between refreshes. That is acceptable only where the application can work with data a little out of date - a catalogue or a reference table rather than a live account balance. The other two options describe conditions in which replication should be avoided.

Answer: A

Correct answer: Unit of work

The recognised stages are remote unit of work, distributed unit of work and distributed request, in order of increasing capability - from a transaction touching one remote site, to one touching several, to a single statement spanning several. A plain unit of work is the centralised case and is not a stage in the evolution at all.

Answer: D

Correct answer: A different DBMS is used at each location and data are distributed across all nodes.

Heterogeneous means the sites run different products, and the difficulty is exactly that: differing SQL dialects, data types, transaction models and catalogue structures must be reconciled before a single query can span them. Homogeneity is easier to manage but is often not available, since sites arrive through acquisition and separate history.

Answer: C

Correct answer: Each site must have the same storage capacity.

A full copy of the database at every site means every site must be able to hold the whole thing, which can be a heavy requirement at a small location. The other three options are advantages of replication rather than disadvantages - less network traffic, resilience to the failure of one site, and local processing without network coordination.