Multi-Model Best Practice
This fundamental explains about sharing data across systems in a multi-model environment using event streaming and web services. It discusses the processes, best practices, and considerations for implementing these concepts. Take a look to gain a deeper understanding of how communication between systems in different models can be achieved effectively.
Should you have any questions, please get in touch with academy@emagiz.com.
1. Prerequisites
- Some context on the eMagiz multi-model environment and integration patterns will be useful:
- Fundamental eMagiz Multi-model
- Fundamental eMagiz Event Streaming
- Fundamental eMagiz API Gateway
- Fundamental eMagiz Messaging
2. Key Concepts
- Model - the integration model of a client that runs on a specific cloud slot of eMagiz
- Multi-model - several integration models of a single client
3. Main Multi-Model Integration Patterns
At the moment, the two most frequently used ways to provide communication between systems in a multi-model environment are event streaming and web services. This microlearning will discuss the processes of sharing data across systems in a multi-model environment using these concepts, including best practices.
In the figure below, both processes are illustrated. The process at the top illustrates event streaming, and the process at the bottom illustrates web services. Model A and model B are two different models, where a system in model B should have access to the same data as the system in model A. For example, the systems in both models should be the same.
3.1 Event Streaming
To understand the following section, it is necessary to have some background information about event streaming, which can be found here.
The first option is to decide the alignment of the data models between both models. Whenever data from the first model is transferred to the second model, and the topic is considered input for other systems and integrations in that other model, the data needs to be transformed to the common data model of that second model. This way, the topic can be exposed to any system and pattern in the second model. The picture below denotes this situation. These are the specific considerations when implementing this approach:
- Leverage the credentials of the first eMagiz model in the second model
- Create transparent systems with clear names and ensure to align the message type names
It is necessary to transform topics to allow communication between systems in different models using event streaming. This is because it is impossible to retrieve data from a topic that exists in one model through a different model. So, systems in model B cannot immediately access the topic present in model A. The topic needs to be transformed to ensure that systems in model B can use the data on the topic in model A. By changing the topic, all systems in model B can access the transformed topic.
To ensure that the topic from model A can be transformed, model B needs credentials from model A. So, to allow for this pattern, credentials must be shared between models.
An advantage of this pattern is that there is no dependency across models. If the system in model A breaks down, the system in model B can still work. A disadvantage of this pattern is that there may be struggles with the setup since no direct accessibility can be achieved. Moreover, as mentioned above, credentials have to be shared, which may threaten security.
3.2 Webservices
Web services can also be used to communicate between systems in different models. This means that model A has a web service in place, and model B has a web service in place, between which data can be exchanged.
An advantage of this pattern is that it is easy to keep an overview. Setting it up does not involve many unclarities. A disadvantage is that there is much dependency across models. If the web service in model A crashes, the system in model B cannot receive data anymore.
3.3 Naming Conventions
For clarity purposes, it is advised to give the systems that are the same across models the same name and the name of the other model in which the system is present. So, as illustrated in the figure, the system names in model A and B are partly the same (‘SYS’) and indicate which model also holds the system (model B in model A, model A in model B). This way, it remains clear which models have the same systems, and systems can be found more easily.
4. Key Takeaways
- Event streaming and webervices are mainly used to communicate between systems across models.
- To allow communication between systems in different models using event streaming, it is needed to transform topics.
- Communication between systems in different models can also directly be achieved using web services.
- For clarity purposes, it is advised to give the systems that are the same across models the same name, as well as the name of the other model in which the system is present.
- Both patterns have their advantages and disadvantages.