Friday, June 25, 2010

Agility: Cloud Benefit Number One

A couple of Team Appoxy attended the GigaOm Structure 2010 conference this week. The big themes that resonated at the conference were agility, big data, and frameworks and orchestration. All things in line with Appoxy’s views but the consistent pattern throughout the two days speaks to consolidation of thinking in the space.

In discussions at different panels on why cloud, agility consistently appeared at the top of the list. Cloud company after cloud company said it was the most vocal benefit they’ve been hearing from their customers. Cost savings, scalability, and rationalization of IT processes certainly rank up there but the most disruptive element for app developers working in the cloud was the agility it gives them to rapidly build applications that benefit customers and drive revenue.


In the airline industry, Southwest gained tremendous efficiencies by using a limited number of airplane configuration (starting at first with the Boeing 737-200 and then progressing through other 737 variants). Using standard configurations allowed them to run a leaner operation and maintain higher aircraft use. (They had to order and store fewer parts, could provide more streamlined training and maintenance procedures, etc.)



Application development in the cloud has the same benefits. Developers and workgroups can make use of consistent stack recipes and server architectures and reduce the number of steps in the process. They can also reduce the number of people involved given the reduction in system administration needs. Add in the use of programming frameworks and data store abstractions and app developers can spend less time on setup, orchestration, configuration, and other system level task and more time creating great applications.

In the airline industry, cost reduction and maximizing airplane use is key. In cloud computing, simplification of development processes translates into increased speed and agility. If you’ve architected and built your application correctly* you’ll be able to scale from pilot to prototype to production without having to worry about moving pieces around or allocating resources for an undetermined need.


Having push-button build-test-deploy processes in place for cloud development (like
SimpleDeployr) gives you a 2x to 3x advantage in app building time from just a few short years ago. For iterating and improving on an app, the benefit is even greater (3x-5x) because teams are able to act more independently and deploy quickly on on-demand servers and connect easily to test datastores, and test routines. This can be for simple back-of-the-envelop work (let me spend an afternoon to try something), as part of formal build-stage-and-deploy development cycles, or anywhere in between.
At the conference, Facebook says they make interface changes at least once a day which indicates they’re running concurrent cycles as well as maintaining some amazingly flexible risk tolerant release practices. The flexibility of the cloud along with good development practices, a/b release tests, and heavy metrics analysis gives them this capability.
If the conference is an indication, you’ll be reading a lot more about agility and the cloud in the coming months. You'll be hearing it in combination with Lean Startup. And it will be talked about in combination with John Boyd's OODA loop. This disruptive aspect isn’t a secret among SaaS companies and social web companies but the dramatic scope of the benefit is still making it’s way across all technology and industry sectors. These sectors know there’s something to the cloud and they know they should be doing things in it but they don’t quite have a measure yet for how to value the cloud. It's late in the game for many that have seen their competitors and upstarts iterate many times faster than them. Hopefully it's not too late for others.

*By running and testing on multiple servers, shutting servers down and sparking them up, and, for big data applications, building a data tier that can more easily scale horizontally.