A well architected Enterprise One system is typically designed and built out to handle peak loads for the business. Often with a little extra to handle busy seasons, future growth, or month-end activities. This is ideal to ensure that the system does not become the bottle neck for the business and employees are able to get their job duties completed.
The downside of this model comes into play when the system is not running at or even near peak loads. Many companies do not have a large volume of employees connected overnight, on weekends or holidays. Although, the system compute overhead is still online and running during these off-peak times. Even consider how often non-production servers are sitting idle during a typical month.
One key advantage to containerization is around the dynamic scaling functionality that can be orchestrated. Looking at the Enterprise One web layer, you can configure one JAS server to be always on. With two production JVMs running it will cover most non-business hour user loads with a degree of redundancy. We configure thresholds on this web container to spawn a clone when the load requires it. Which allows the system to dynamically grow as more users connect. Low end thresholds are also configured to allow the system to reduce the footprint as user loads drop.
Expanding the Enterprise One system to handle company growth can be costly to a budget. What is the typical cost and time to install and configure an additional Enterprise One server to the system? With the container architecture, we adjust a few container settings for that server type and ensure we have the container engine base to handle the additional load.
Dynamic architecture can provide intelligent infrastructure and application elasticity to your Enterprise One system. Which in turn will provide a reduction in overall compute costs for the system. We will delve into additional features that containerization will provide on our next post in the series.
- Todd Burton is a Senior JD Edwards Technical Architect with Main Street APPs.