Article Preview
TopIntroduction
In today’s digital era, there are countless real-world applications requiring access to the remotely distributed data present at multiple distant sites (Omamo, Rodrigues, & Muliaro, 2020) (M., K., & K., 2020). Telecommunication services and online trading systems are some of the easy-to-understand example applications (Mustafa, 2021) (Gupta & Shanker, 2020). Here, data is dispersed across geographically distant sites and a wide-area network is utilized to communicate among these sites (Pandey & Shanker, 2016) (Srivastava, Shankar, & Tiwari, A protocol for concurrency control in real-time replicated databases system, 2012). All such applications exhibit characteristics like a fast exchange rate with data access from anywhere at any time (Pierce, Shepherd, & Johnson, 2019) (Kizito & Semwanga, 2020). Besides, the location of a node/device can also be useful for some custom applications (Gupta & Shanker, 2021). With increased application complexity, the underlying data and transactions accessing that data both are associated with time constraints or deadlines. In simple words, one needs to take care of the fact that the data deadline is also honored in addition to the transaction deadline (Ulusoy, 1994) (Singh, Pandey, & Shanker, 2019). These applications require data that is constantly changing — weather data, stock prices, health indicators etc. So, data generated by such applications have some validity associated with it. This means that data can be used to take some action using transaction only before the expiry of its validity.
The above-discussed applications are studied under the umbrella of a research area named “Distributed Real-time Database Systems (DRTDBS)”. The DRTDBS is a finite set of geographically separated database sites connected through a network (Pandey & Shanker, 2020) (Shanker, Misra, & Sarje, 2006) (Pandey & Shanker, 2018). Each database site consists of one local database. Considering the whole set up as one logical unit, any such change in the state of data could be performed using real-time distributed transactions. A real-time distributed transaction is a real-time transaction that requires access to both local and remote data items. Going one level further down from a definition perspective, a real-time transaction is a transaction with associated time-constraint/deadline. The two key issues with DRTDBS based applications are deadline miss of transactions and expiry of data deadline. These issues mainly occur because of longer data access latency. A transaction processing framework mainly consists of the following components — priority assignment scheme (Pandey & Shanker, 2019), concurrency control (Pandey & Shanker, June 18-20, 2018) (Pandey & Shanker, 2017), and commit processing (Pandey & Shanker, 2018) (Pandey & Shanker, 2019). Replication of data requires adjustment in all the above 3 components.