The IBM PureApplication System is a great platform to rapidly deliver applications to users. Traditional Java web applications can be easily deployed and scaled to match demand.
Applications for open enrollment, employee charitable contribution campaigns and employee surveys, among others, are examples of applications that have defined peak periods of usage and then sit near idle when not in use. Do you design and deploy in preparation for the peak? Or do you deploy to a platform that has elasticity beyond the ability to increase CPU and memory to a single virtual machine? It is important not to confuse platform (virtualization) elasticity with application elasticity.
Let’s consider a simple web application. The screenshot below illustrates how we can build a pattern that leverages an enterprise Active Directory, a DB2 database and a WebSphere Application Server tier, which we can configure to scale automatically based on demand and deprovision resources as the utilization decreases. We can do this with the PureApplication System automatically, with minimal configuration.
The application topology is simple enough—we have all seen it before. Note the scaling policy icon. We added this to the pattern; it will allow us to apply and configure a scaling policy to the web application.
We can now configure the scaling policy for this application.
For scaling type we selected “Response Time Based”; however there are multiple scaling types. You can select Static, CPU Based, Response Time Based or Web to DB. For response-time-based scaling we can configure the web response time threshold range in milliseconds and the minimum time (in seconds) to trigger add or remove. This specifies the time duration condition to start scaling activity.
Once this pattern is deployed this policy will be enforced to allow scaling in and out dynamically based on response time. No additional configuration is required, and in addition we can monitor the application through the PureApplication System console to understand application performance.
The IBM PureApplication System and the virtual application patterns can simplify application scaling strategies with built-in expertise and minimal configuration. For applications that have high peaks but a short shelf life, it allows you to scale as demand grows without deploying infrastructure for the peak.