ServiceSim: A Modelling and Simulation Toolkit of Microservice Systems

ServiceSim is a toolkit built on the CloudSim simulation framework. It is designed to simulate the running of microservice systems in large-scale, distributed cloud-edge environments.

Main features

• support for modeling and simulation of large scale cloud-edge computing environments
• support for modeling and simulation of microservice systems, including service registration and discovery, load balancing, load admission, request dispatching and invocation relationship between microservices.
• support for verification of various microservices system configuration policies, including service deployment, request dispatching, load balancing, and load admission, among others.

Download

The downloaded package contains all the source code, examples and jars.

Publications

• Shi H, He X, Wang T, et al. ServiceSim: A Modelling and Simulation Toolkit of Microservice Systems in Cloud-Edge Environment[C]//International Conference on Service-Oriented Computing. Cham: Springer Nature Switzerland, 2023: 258-272.

Test Examples

• TestExample Test the processing of end-user requests generated from files on two edge nodes
• TestExample1 Test the processing of randomly generated end-user requests on 32 edge nodes
• TestExample2 Test the processing of randomly generated end-user requests on 32 edge nodes, and the deployment of microservices is different from TestExample1
• CompK8STestExample2 The first test example compared with K8S. Test the processing of end-user requests that arrive at the system at specific times.
• CompK8STestExample3 The second test example compared with K8S. Test the processing of end-user requests that arrive at the server where the first service is located at a specific time.
• CompAllUpdateTestExample1 The third test example compared with K8S. Compared to the configuration of CompK8STestExample3, there are differences in the infrastructure setup.

Explanation: Partial files of test results have been retained in the project. Before running the above test examples, please delete these test result files or name them with a different name and generate a new test result file. Please refer to the test examples code for specific details.

Configuration for Starting Simulation

To create a new test example and start simulation, the following configuration is required:

• DevicesProvider You need to implement the interface class DevicesProvider. Implement this class to configure the cloud-edge environment. The project provides two simple implementations of DevicesProvider, namely DevicesProviderSimple and DevicesProviderSimple1. In the examples, initialization of the cloud-edge environment is required, including the nodes (NetworkDevice) and the network connections between the nodes (Channel).

• ServiceProvider Once the cloud-edge environment is configured, the setup of the microservices system needs to be completed. You need to initialize a ServiceProvider class. The configuration includes the following:

  • Servicechain The invocation relationships between microservices are represented by ServiceStage. Refer to our publication or NetworkCloudSim for details.
  • LoadAdmission You can implement this interface class to design the admission conditions for end-user requests. The NonLoadAdmission has been implemented in the project to allow all end-user requests to enter the system for processing by default.
  • LoadBalance You can implement this interface class to design the load balancing mechanism for each node. The RoundRobin method has already been implemented in the project.
  • RequestDispatchingRule It holds the service discovery information for request dispatching between nodes. By implementing this class, you can control the scope and prioritize of request dispatching. The simple request dispatching priority rules are implemented in RequestDispatchingSimple.
  • initInstance The initial service deployment situation also needs to be configured.
  • cloudletResultFile Writing the results of the execution to files, and setting the filenames is also necessary.

• EndUser This class implements various methods for generating end-user requests, including reading from files and random generation. You can simulate the arrival of end-user requests to the system using these methods.

To initialize a new simulation instance, you need to complete the above configurations. After that, you can start running the simulation. The previous section introduced six test examples that can be used as references.

Runtime Environment and Tool Recommendations

• Java JDK 1.8
• Intellij IDEA