Business use

As described in Introduction to hots, hots is meant to handle time series based environments and optimize these time series management. We tried to remain as generic and as adaptable as possible, but hots has been developped dealing with the specific containers resource allocation context. This genericity effort can be seen through two main parts, which can be easily adapted by users to tackle their specific use case.

Optimization models modularity

In order to evaluate the current solutions, hots use some optimization models to find the way to update these solutions (more details in Pyomo use). In the package, the currently used optimization models are described in the model.py and tackle two problems :

• the clustering problem

• the containers placement problem, using clustering information

The user has the possibility to provide his own file (or modifying the provided one), in order to adapt the optimization constraints, variables and / or objectives to his use case, being related to containers placement or not.

Business components modularity

Aside from the optimization part, the functions developped for handling the business use case (in our case the container resource allocation) are grouped in a specific file. For example, in the analysis period, several heuristics have been developped in order to propose a first containers placement solution with first historical data : these heuristics are found in the placement.py file.

Besides the placement problem, we wanted to tackle the problem of resources allocated to containers : a new file allocation.py has been created and linked to the main.py file. It shows the possibility for the user to use either alternatives for heuristics developped in hots, or one full new use case, defining the algorithms and the optimization models use.