Developing An Intelligent Infrastructure For Transaction Management

Published Date: 02 Nov 2017

Abstract— Grid architecture is a fast growing distributed computing Environment. Although various distributed computing architectures were proposed, yet concurrency control is a challenging issue in some part of research. Considering the ACID rules while performing atomic transactions creates high complexities in grid systems. Due to the complex social structure and highly mobile nature of the life, there is currently a need for developing the technology for a large scale, highly distributed, self-adaptable and service-based mobile system. To resolve this problem, this paper introduces a protocol called C3G (Concurrency and Consistency Controlling in Grid system) which can deal with consistency, concurrence, atomicity, distribution and large volume of data in Grids. The correctness of the protocol was examined followed by performance valuation through simulation and mathematical analysis.

Keywords: Concurrency Control, ACID, Transaction, Grid Database, Data Grid.

Introduction

Transaction management has one important feature called concurrency control to ensure consistency of the database. In mobile computing environment, data management becomes a challenging issue due its limitations i.e. variable bandwidth, frequent disconnections, limited resources on mobile host etc. Several valuable techniques are proposed in literature to provide concurrency control in mobile environments; however most of them are based on locking, time stamp and optimistic concurrency control.

Applications such as computation chemistry, climate forecast, tele-medicine, multiphysics, and other scientific applications have to manage large-scale distributed data. These applications gather gigabytes of data per day. Also data from these applications are best stored locally but they will be required to access globally for research purpose. Under these conditions local databases will evolve autonomously and the nature of underlying management systems may also be completely heterogeneous. This leads to the need for developing the technology for a large scale, highly distributed, self-adaptable and service-based mobile system.

Considering the recent technological advances in the area of the Grid and network reconfigurability, it is now feasible to develop such a universally distributed and dynamic system, which in turn provides competitive opportunities for the value-added service providers. Traditionally, the value-added service providers have limited options for providing adaptable services due to the available resources and inflexibility of the network itself.

A generic solution can be envisaged integrating communication network and Grid system to suit the mobile applications. It is believed that the Grid may form the basis for a universal infrastructure to serve any type of terminals anywhere and anytime providing the required computer-based resources.

To resolve this problem, this paper introduces a protocol called C3G (Concurrency and Consistency Controlling in Grid system) which can deal with consistency, concurrence, atomicity, distribution and large volume of data in Grids.

RELATED WORK

In [2] mobile database environment, providing consistency of the data items is a challenging issue in case of concurrent access. Various valuable attempts are made in providing solutions for data management in mobile environments. The conventional two phase locking protocol is not suitable, as it requires clients to communicate continuously with the server to obtain locks and detect the conflicts.

A distributed OCC (Optimistic Concurrency Control) method followed by locking, such that locking is an integral part of distributed validation and two-phase commit was discussed[10]. This method ensures at most one re-execution, if the validation for the optimistic phase fails. Deadlocks, which are possible with 2PL, are prevented by preclaiming locks for the second execution phase. This is done in the same order at all nodes. For higher data contention levels, the hybrid OCC method allows a much higher maximum transaction throughput than distributed 2PL in systems with high processing capacities. The protocol is fully distributed and employs parallel validation and lock acquisition.

A hierarchical transaction model for the execution of distributed transactions with mobile code on open networks was proposed[5]. The presented transaction model is an open nested transaction model. This model supports those parts of a distributed transaction which is executed asynchronously in relation to other parts of the same global transaction. Furthermore, the model is able to recover the execution of a transaction when a sub-transaction of this transaction becomes unavailable for a long period of time.

The grid real-time transaction model based on function equivalent service is defined [4]. The model supports the joint implementation by a number of function equivalent services which own the same transaction, and allows grid services to join or exit the transaction’s execution dynamically. The Share and Compete (S&C) concurrency control protocol solves resource conflicts between function equivalent services of the internal transaction in the grid real-time transaction model based on function equivalent service. The basic idea of S&C protocol is it Share Resources.

A Multi-scheduler concurrency control protocol was proposed to meet the requirements of Grid databases [7]. The Grid Concurrency Control (GCC) protocol has two major phases: (a) submission phase and (b) termination phase.

Transaction management in Mobile Database System is more complex because of unlimited mobility of the Mobile Unit (MU). A mobile-agent-based framework was proposed to facilitate efficient transaction processing during handoff or MU failure[6]. Instead of executing transaction in the mobile unit it is given to base agent in the base station. Speculative locking (SL) protocol was used to improve the performance of fixed networks. The SL protocol with agents reduces the waiting time of connected mobile unit and facilitates efficient transaction processing during handoff and mobile unit failure.

Different transaction management techniques and protocols were designed and implemented to maintain concurrency control in the Grid and Mobile environments. But when a mobile host initiates a transaction in the Grid environment it has certain limitations such as variable bandwidth, frequent disconnection, and limited resources on mobile host. These are the issues that are taken into consideration while designing the proposed model.

PROPOSED C3G MODEL ARCHITECTURE

A transaction is a collection of operations on the application state. A transaction has several important properties, including the ACID properties maintaining the application in a consistent and durable state.

The management of transactions is the responsibility of the Transaction Manager. The Transaction Manager coordinates the application of transactions to Resource Managers, which hold the data and other objects whose state is to be changed by the transaction.

The two laws of concurrency control are:

Concurrent execution should not cause application programs to malfunction.

Concurrent execution should not have lower throughput or much higher response times than serial execution.

Though many techniques and approaches have been designed and implemented for maintaining consistency of the TP system, most of the techniques uses rollback as a recovery for failed transaction. In the case of mobile environment the transaction may fail due to the frequent disconnection of the mobile host, varying bandwidth, limited battery power etc. In that case it may lead to more number of transaction failures and restarts thus affecting the efficient performance of the system.

The proposed model consists of an agent which runs the concurrency control protocol C3G (Concurrency and Consistency controlling in Grid system). When the mobile host initiates a transaction the agent triggers the protocol. The protocol maintains the consistency of the system.

The C3G model architecture is given in the fig 1. A user interaction operation is issued by a mobile host during the execution of a transaction generated by it. This operation can be considered as reading a local data item that is required to continue the execution of the mobile transaction. Therefore, a user interaction request is satisfied locally, and there is no need for wireless link communication. However, a wireless link delay is experienced between the generating mobile host and the current MSS for satisfying the user interaction requirement of the mobile transaction. The request is outsourced to server present in another network. A steeper degradation in the performance is observed as the number of user interactions increases.

Fig 1 C3G Model Architecture

A mobile agent is a process that can transport its state from one environment to another, with its data intact, and be capable of performing appropriately in the new environment. Mobile agents decide when and where to move. Movement is often evolved from RPC methods. Just as a user directs an Internet browser to "visit" a website (the browser merely downloads a copy of the site or one version of it in the case of dynamic web sites), similarly, a mobile agent accomplishes a move through data duplication. When a mobile agent decides to move, it saves its own state, transports this saved state to the new host, and resumes execution from the saved state. A mobile agent is a specific form of mobile code, within the field of code mobility. However, in contrast to the Remote evaluation and Code on demand programming paradigms, mobile agents are active in that they can choose to migrate between computers at any time during their execution.

Serializability is ensured through the adaptation of the Three phase locking scheme. Priority Abort protocol is used for resolving data conflicts (i.e. a low priority transaction is aborted when one of its locks is requested in a conflicting mode by a high priority transaction). Each fixed site in the system has a scheduler to manage the lock requests of the cohort processes executing at that site. Each cohort process has to obtain a shared lock on each data item it reads, and an exclusive lock on each data item it writes.

Consistency of a transaction is maintained by using a modified version of concurrency control protocol. This modified version, adapts C3G to a mobile computing environment. C3G is designated as the coordinator, and each cohort process executing on a data site acts as a participant. When a transaction commits, the updates of its cohort processes are stored in the local databases.

Due to the mobility of users, mobile hosts may cross the boundary between any two cells. In order to keep the connectivity of a mobile host to the fixed network a handoff process is implemented. During handoff, the new cell’s MSS takes the responsibility of providing a wireless interface to the mobile host. This process should be transparent to the user. We assume that each MSS broadcasts beacons over its wireless link. Each beacon carries its sender MSS’s address. A mobile host monitors the wireless signal strength it receives from neighboring MSSs. The mobile host may decide to initiate a handoff process when the signal received from a new MSS is substantially stronger than that received from the current one.

Initially the mobile device starts a transaction. The transaction is coordinated by the Transaction Manager. The transaction is carried out, if it succeeds, then the values are updated in the database and the transaction is committed. In case of failure the values are rolled back. If there is problem in the transaction due to variable bandwidth, disconnection or limited battery power then the concurrency control protocol C3G is triggered. The protocol takes the backup of the transaction and stores in a temporary storage. When the connection is restored the transaction is resumed from the point of interruption. The flow is depicted in the figure 2 given below. A mobile data grid system that supports a real

Mobile Transaction

Mobile Base Station

Outside network

Request Response Protocol

Transaction Manager

Concurrency Control Protocol

Commit

Rollback

Database

Transaction Fails

Temporary Store

Fig 2 Data Flow Diagram of Proposed Model

time response to transactions of underlying applications was designed. Three important factors concurrency, consistency and atomicity were taken into account. The transaction operation on the granted data is executed at the mobile host. Host request is transferred to fixed part of other network. If any transition fails then the transaction state would be maintained in the Agent and would effectively manage database and mobile client information on server side.

EXPECTED OUTCOME

An agent based model C3G is designed. The proposed model will maintain the consistency of the database in the Mobile Data Grid. The model improves the performance of the system. Also efficiently reduces the concurrency problem in mobile transaction. The following are the expected outcomes of the proposed model

Improved Performance

Increased Throughput

CONCLUSION

Introduced a protocol called C3G (Concurrency and Consistency controlling in Grid system) which can deal with consistency, concurrency, Atomicity, distribution and large volume of data in Grids. A mobile data grid system that supports a real time response to transactions of underlying applications was designed. Three important factors concurrency, consistency and atomicity were taken into account. The transaction operation on the granted data is executed at the mobile host. Host request is transferred to fixed part of other network. If any transition fails then the transaction state would be maintained in the Agent and would effectively manage database and mobile client information on server side.