Big news everyone. Developing and deploying enterprise applications costs money, and often lots of it. For once, I’m not going to make claims about the huge sums you can save by deploying F5, but do I think there is an interesting discussion point about where you spend the budget for a new application.
A few times lately I’ve seen customers consider lower priced alternatives to F5 BIG-IP for providing load balancing services, due to budget constraints. I’m not going to rubbish these competitors, many do a reasonable job at a price point. When you look into it, however, it soon becomes clear that there is a balance between spending less on the Application Delivery Controller (ADC) layer and more on the application development and associated platforms. These costs also extend into future application projects since they will frequently end up sharing the same functionality-limited delivery platform. Removing capability from the ADC shifts the requirements down the stack, leaving the application code to fill the gap.
That’s all well and good apart from two key issues:
1) Functions developed for an application tend to stay in that application and are not simple to reuse quickly for other applications. Functionality is essentially ‘locked’ into the code.
2) Some application functions are best done at a higher point in the stack, rather than repeated on multiple application servers. Things like decryption, localisation, authorisation and workload management (think queuing systems, cloudbursting etc.) just work better at the ADC layer where it’s position as a strategic point of control can be used to full advantage.
So my advice is this: don’t lock all the functionality inside your application - once you set it free, you can reuse it time and time again, cutting the cost and time to deliver the next one.