Legacy IT operations management (ITOM) solutions tend to ignore the importance of a real-time IT infrastructure model that spans cloud services all the way down to a chassis fan. On the other hand, one of Zenoss’ key tenets is model-based monitoring, which provides the context for interpreting the data coming from the environment and allows more intelligence to be derived for faster and automated decision making. In a previous post, the basics of what a model comprises and how the complexity grows with size were explained, and in this post, the methods for building the Zenoss model as well as the model composition are explained.
Creating the Zenoss model depends upon ZenPacks, which are plug-ins, making Zenoss aware of the different topology entities within the IT environment, whether physical or virtual. ZenPacks provide much more than just the model metadata, as they contain domain-specific information informing the management system of collection mechanisms, event classes, thresholds, user interface (UI) enhancements, etc., but for the purposes of this conversation, the focus is on the model-driven aspects. ZenPacks work in tandem to assemble a holistic real-time view of the entire IT infrastructure, and the elegant design of the ZenPack SDK ensures that the Zenoss IT infrastructure model can be easily augmented and modified without limitations for the future.
The metadata contained within the model is derived by interrogating the environment and piecing together a complex web of interconnected services and components. Here’s a small sampling of gathered information representing how relationships within the model are created.
- IP addresses (Layer 3)
- MAC addresses (Layer 2)
- Storage addresses (WWNs, IQNs, etc.)
- Management APIs that reveal detailed model information. (vSphere API, UCS API, etc.)
- Logical metadata (VM IDs, tags, etc.)
What makes IT infrastructure model data more interesting relative to other types of data is that these relationships are not just hierarchical in nature, and, in fact, the model is akin to a graph data structure where a node may, and often will, participate in multiple edges. Zenoss has acquired decades of domain knowledge to understand what relationships are important and then utilizes special technologies to rapidly traverse these complex models. Since Zenoss automatically creates and maintains the entire IT infrastructure model in real time, users are able to use this knowledge for higher-level decision making with regard to events and to understand the impact relationships of the infrastructure as a whole instead of a piecemeal approach.
Managing all of this model information across hundreds of thousands of nodes across data centers and cloud environments quickly becomes overwhelming. Zenoss tackles this problem by providing templates and preformed classification mechanisms, allowing a few changes to be impactful across a larger number of services and components. As a slight teaser for the future, upcoming Zenoss products will include the notion of model informed as well as the ability to create the model based on machine learning, which will further augmentation of the model into areas where connections within the model are not precisely determined. In upcoming posts, we’ll discuss some of the key uses of the model within the ITOM landscape and delve into alternative perspectives of the model that provide intelligence above legacy monitoring solutions.
