Adaptive Lockable Units to Improve Data Availability in a Distributed Database System

Distributed database systems have become a phenomenon and have been considered a crucial source of information for numerous users. Users with different jobs are using such systems locally or via the Internet to meet their professional requirements. Distributed database systems consist of a number of sites connected over a computer network. Each site deals with its own database and interacts with other sites as needed. Data replication in these systems is considered a key factor in improving data availability. However, it may affect system performance when most of the transactions that access the data contain write or a mix of read and write operations because of exclusive locks and update propagation. This research proposes a new adaptive approach for increasing the availability of data contained in a distributed database system. The proposed approach suggests a new lockable unit by increasing the database hierarchy tree by one level to include attributes as lockable units instead of the entire row. This technique may allow several transactions to access the database row simultaneously by utilizing some attributes and keeping others available for other transactions. Data in a distributed database system can be accessed locally or remotely by a distributed transaction, with each distributed transaction decomposed into several subtransactions called participants or agents. These agents access the data at multiple sites and must guarantee that any changes to the data must be committed in order to complete the main transaction. The experimental results show that using attributelevel locking will increase data availability, reliability, and throughput, as well as enhance overall system performance. Moreover, it will increase the overhead of managing such a large number of locks, which will be managed according to the qualification of the query. Keywords—Granularity hierarchy tree; Lockable unit; Locks; Attribute level; Concurrency control; Data availability; Replication