I frequently joke with my team about finding our next customers in major business publications. Every time I read a headline like the following, I immediately think: “Those guys need LISA!"
Yahoo Tech News: "Apple, AT&T sites buckle as iPhone 4 pre-orders begin"The fact is, all the agile developers, functional testers and load testing servers they can fire up are not going to help them here. This is NOT a load testing issue, it’s a virtualization of services issue!
If we met the guys working on all this critical stuff, you would certainly not hear that companies like Apple or AT&T don’t have enough money to buy servers. In fact, many major companies actually tell me that they have more servers registered in their asset tracking system than they have employees! These guys will be no different...
So how could these teams have used LISA to prevent this conundrum?
- Service Virtualize each others' pieces of functionality that contribute to the site experience, so that each team can develop and functionally test their bits, without having to wait for the system to come together later, during integration.
- Service Virtualize the dependent systems (inventory, customer ordering, account and number provisioning, etc.) so that performance tests at high levels can be performed ahead of time, and each service can size and qualify their environment for the load expected when it goes live.
Some day in the future, they will even want to have intelligent ways to fail over in production during peak load hours, but let's not put the cart before the horse. One of my banking customers puts it eloquently when he tells his component teams to reach a certain performance level on their own: "If your app isn't performing fast enough in your own virtual environment, why should you bother giving it to me to test in my environment?"
OK I’ll stop rambling. Now somebody let me know when the coast is clear, so I can order my next phone ... ;)
