Database Reference
In-Depth Information
Replication Transparency
You know that the PRODUCT relation is replicated and that full copies are stored
in three sites. Assume that a user in Houston wants a product list. When the query
is submitted, the user has no knowledge of whether there is one copy or more copies
of product. In fact, the user need not be concerned. The DDBMS hides the details
of replication from the user and provides replication transparency.
The DDBMS knows that there is no local copy of product data in Houston. So
it selects a copy from one of the other three sites, enables the query to execute, and
produces the result. When there are copies at multiple sites, the DDBMS chooses
the best possible candidate that will execute the query efficiently. This takes in con-
sideration network traffic and other factors. If one of the sites, namely, Los Angeles,
is not operational at that time, the DDBMS will go to other sites even though they
are further away by comparison.
Location Transparency
Through the property of location transparency, the DDBMS hides from the user the
details of where parts of the database are stored. To the user, it appears as though
all data are stored at the local site. How many sites there are and how the database
is spread to any sites—these are totally transparent to the user.
For example, assume that the Chicago site is given the responsibility of analyz-
ing all the orders for the past six months. A user in Chicago charged with this
responsibility submits a query to retrieve selected data from the entire ORDER
and CUSTOMER relations. You know that these two relations are partitioned
horizontally and the data are spread across all sites. However, the query from the
Chicago user need not be concerned about where all of these data are coming from.
The DDBMS hides all the distribution details from the query. The coding of the
query is exactly as though it is coded for a centralized database. The DDBMS
enables data to be retrieved from all the partitions at different sites and consoli-
dated to produce the result.
Network Transparency
In an earlier subsection, we discussed the various methods of configuring the
communications network. We looked at LANs and WANs. We mentioned fully
connected and partially connected networks. We also considered a few topologies
such as bus, star, and ring. Now suppose that a user in New York needs to create a
confidential employee report printing full details. Should the user be concerned
whether New York and any other site that may store employee data are connected
directly? Should the network configuration affect the way the user query must be
coded?
The DDBMS makes the data communications network, its protocols, and its
configuration transparent to the user. The user does not even need to be concerned
whether there is a network at all. The user's query will be coded as though
all data will be retrieved locally without any complex networks to support the
retrieval.
Search WWH ::
Custom Search