Our developers Frederick Proske and Robin Schuchmann will present their work at this years EURO conference in Espoo, Finland:
Typically, personal computers have been powerful enough to quickly solve the model instances generated by GAMS. If not, users (or their expert IT staff) have implemented custom scheduling systems to run large optimization jobs on central compute resources. Increasingly, users want to run large jobs or large streams of jobs in the cloud. This allows them to access more powerful computers than the usual desktop computers and also to use a highly scalable pool of worker machines.
In this talk, we present GAMS Engine - a cloud-native solution for executing long-running (optimization) jobs on a highly flexible pool of heterogeneous nodes. We describe how we use Kubernetes to develop such a system for clusters of any cloud provider or on-premises, and the particular challenges optimization jobs pose to the design of such a cloud native system.
We will also talk about how GAMS Engine enables customers to fully automate their OR-related business processes and scale in real time as needed, both vertically (e.g., from a few MB to several TB of memory) and horizontally (from a single job to many parallel scenarios).
The evolution in software and hardware has a great impact in the field of mathematical optimization. Apart from obvious benefits such as increased computing power, the way software is used has also changed significantly. Whereas optimization software used to be invoked via the command line on a local computer, today many users want to log into an online service and perform their optimization on a centralized compute resource. Another tendency that has been emerging for some time is that more and more software, as well as optimization models, are being made available as open source.
These and other trends do not stop at GAMS. In recent years, various new developments have been initiated to meet the changing requirements. In this talk, real-life examples are used to show what modern software solutions with GAMS can look like. Different approaches will be discussed, from the integration of a cloud-based GAMS system into existing infrastructures to entirely redesigned applications whose graphical user interface can be operated via a smartphone. We will also discuss how to organize projects in an efficient way and how to facilitate communication with the customer.