With the emergence of Cyber-Physical Systems (CPS), several sophisticated runtime monitoring solutions have been proposed in order to deal with extensive execution logs. One promising development in this direction is the integration of time series databases (TSDBs) that support the storage of massive amounts of historical data as well as providing fast query capabilities to reason about runtime properties of such CPS.
In previous work, we presented TemporalEMF, which added native temporal support for models. In TemporalEMF, models are automatically treated as temporal models and can be subjected to temporal queries to retrieve the model contents at different points in time. For that, we were using a NoSQL database as a backend. We are now going one step further and explore how conceptual modeling can benefit from TSDBs, and vice versa. By using TSBDs as storage mechanism we can provide more advanced time-related analysis of the modeling data.
The aim is to ensure the traceability between design and simulation/runtime activities by retrieving and accessing runtime information as time series data in design models. In particular, we present how metamodels and their instances, i.e., models, can be partially mapped to TSDBs. For this purpose, we present a profile (cf. Figure below) to annotate metamodels in order to automatically generate wrappers to time-series databases that enable storing model updates as well as querying historical model information.
This profile is part of our paper: Temporal Models on Time Series Databases by Alexandra Mazak, Sabine Wolny, Abel Gómez, Jordi Cabot, Manuel Wimmer, and Gerti Kappel. The full paper is available at the JoT website (open access!). In the work, we explore the interrelation of models and time-series databases around these four main topics:
- We introduce a dedicated time series profile for annotating design models,
- We present a mapping strategy for integrating a metamodel framework (EMF) with a TSDB (InfluxDB)
- We demonstrate how continuous time series queries can be combined with the Object Constraint Language (OCL) for navigation purpose through models, and
- We perform an initial evaluation regarding storage and query performance.
We present two different mapping strategies in terms of feasibility and scalability. While the current work presents interesting insights how modeling technologies may be combined with TSDBs, we foresee additional lines of research worth investigating in future work. Briefly outlined: on the modeling side, we need to deal with co-evolution issues given that the TSDB is schema-less. On the mapping side, we will investigate how to run approximate queries to deal with a variety of uncertainty scenarios as well as study the potential of combining both, temporal and time-series information.