Computational Analysis of .NET Remoting and Mobile agent in Distributed Environment

A mobile agent is a program that is not bound to the system on which it began execution, but rather travels amongst the hosts in the network with its code and current execution state (i.e. Distributed Environment).The implementation of distributed applications can be based on a multiplicity of technologies, e.g. plain sockets, Remote Procedure Call (RPC), Remote Method Invocation (RMI), Java Message Service (JMS), .NET Remoting, or Web Services. These technologies differ widely in complexity, interoperability, standardization, and ease of use. The Mobile Agent technology is emerging as an alternative to build a smart generation of highly distributed systems. In this work, we investigate the performance aspect of agent-based technologies for information retrieval. We present a comparative performance evaluation model of Mobile Agents versus .Net remoting by means of an analytical approach. A quantitative measurements are performed to compare .Net remoting and mobile agents using communication time, code size (agent code), Data size, number of node as performance parameters in this research work. The results depict that Mobile Agent paradigm offers a superior performance compared to .Net remoting paradigm, offers fast computational speed; procure lower invocation cost by making local invocations instead of remote invocations over the network, thereby reducing network bandwidth.

💡 Research Summary

The paper presents a comparative performance study of two distributed computing paradigms: .NET Remoting and Mobile Agents. The authors begin by outlining the landscape of distributed application technologies, ranging from low‑level sockets to high‑level web services, and note that while many of these mechanisms (RPC, RMI, JMS, etc.) are well‑established, Mobile Agent technology is an emerging alternative that promises to reduce network traffic by moving code and execution state across hosts.

To evaluate the two approaches, the authors define four quantitative metrics: (1) communication time (the round‑trip latency for a request/response), (2) code size (the byte count of the agent or remote stub transferred), (3) data size (the payload transmitted for each operation), and (4) number of nodes involved in the computation. They construct an analytical model that decomposes total execution time into communication overhead, local processing time, and additional overhead such as serialization/deserialization.

The experimental methodology replicates a realistic workload consisting of file‑search, data‑aggregation, and simple arithmetic tasks. Identical workloads are executed under both .NET Remoting (using C# MarshalByRefObject) and a Java‑based Mobile Agent framework (Aglet). Tests are performed on two network configurations—a high‑speed LAN (≈10 Mbps) and a constrained WAN (≈1 Mbps)—and the number of participating nodes is varied (2, 5, 10, 20). For each configuration, the authors systematically vary request/response sizes to observe scaling behavior.

Results consistently show that Mobile Agents outperform .NET Remoting across all measured dimensions. The primary advantage stems from the “one‑time migration” characteristic of agents: after an initial transfer of the agent’s code and state, subsequent operations are performed locally on each host, eliminating repeated remote method invocations. Consequently, communication time grows only modestly with node count, whereas .NET Remoting’s latency increases linearly with the number of remote calls. Even when the agent’s code size is relatively large, the one‑off transfer cost is amortized over many local executions, leading to lower overall execution times. Data size also has a reduced impact on the agent approach because the bulk of the payload resides on the host after migration, while .NET Remoting must repeatedly transmit the same data across the network.

The authors acknowledge several limitations. Security concerns are highlighted: mobile agents introduce attack surfaces such as code tampering and malicious payload propagation, which are not addressed in the current study. The cost of serializing and deserializing the agent’s execution state, especially for complex objects, is omitted from the performance model and could become significant in real‑world scenarios. Moreover, the .NET Remoting experiments do not incorporate recent optimizations like asynchronous calls, message compression, or client‑side caching, potentially biasing the comparison in favor of agents.

In conclusion, the research demonstrates that, for information‑retrieval‑type workloads in distributed environments, Mobile Agent technology can achieve superior performance relative to traditional .NET Remoting. The benefits are most pronounced in settings with many nodes, limited bandwidth, or high call frequencies, where reducing network round‑trips yields tangible speedups. The paper suggests future work on strengthening agent security, refining state management overhead, and extending the evaluation to heterogeneous platforms and more diverse application domains.