Oracle’s Unbreakable Linux Network (ULN) is a database application. It is the application that people who have purchased support for Oracle Linux and Oracle VM can use for obtaining rpms, subscribing to non-default channels, downloading errata, requesting Ksplice access, and managing multiple registered servers and CSIs while being guaranteed a certain level of support when and if things go wrong. This application, like many other database applications, has been built with multiple parts, both new and legacy. It is an amalgamation of C, Python, Java, PL/SQL and APEX using Oracle Linux, because using each helps accomplish something that assists the overall application in the long term. Does this sound like a familiar scenario? If so, you are in a common situation. And if not, you are probably one of those unfortunate IT staff members caught in a programmatic ideological war that serves the needs of the programmers more than that it serves the purpose of the actual application(s) currently being planned or worked upon.
This presentation is not so much about what ULN is or does, as an application, so much as it is about what each developer and DBA goes through when building and deploying a multi-piece application that must be maintainable, scalable, and globally available on a 24x7 basis.
When you are tasked to create an application the questions you should ask yourself are: How quickly can I reasonably build this application? How simply? How can I ensure maintainability? How can I assist scalability? How can I best serve the needs of my users in terms of reduced downtime? How flexible should I be towards different ways of accomplishing various development tasks? What is my budget and set of resources? And (perhaps, most importantly) when should I opt not to reinvent the wheel?
This presentation provides a case study overview using the ULN application through lecture and sample code. Topics to be discussed are:
• What legacy code was inherited and what was done with it? Why?
• In what ways are the multiple selected programming languages used to accomplish various ULN tasks? What problems do they avoid and what processes do they assist?
• How is the constant worry of seeking reduced downtime made easier?
• What sets of choices can be made given budget, resource and time restrictions (all rolled into the above)?
• How do you not allow these choices to ultimately impede sound application design?
This presentation addresses these questions and provides a case study overview through lecture and sample code.
