At fnSys we have extensive experience not only developing systems from scratch, but also taking the projects through to the deployment and maintenance phases in some of the most challenging technical environments. Our experience with these systems has  led us to focus on three areas when bringing a system to production-ready status: reliability, scalability, and maintainability.

RELIABILITY

Success in software projects is defined by uptime.  In the 1980s and 1990s most Telco customers demanded "five nines" of reliability in their systems (99.999% uptime, equivalent to 15 minutes of unscheduled downtime per year).   While the cost associated with delivering this level of reliability is prohibitive for most companies, the lessons learned in those demanding environments can be applied judiciously (and cost effectively) to ensure the stability of a platform or service today. 

SCALABILITY

With the advent of the internet and Web 2.0, it is possible for startups to be too successful.  They attract so many users that their platform collapses under the strain.  If the original idea is good enough the door opens for competitors that can deliver the service reliably.  One poor architectural decision wastes all the hard work spent defining and developing a groundbreaking product.  Good system designs need to allow for one hundred users, or one thousand, or one million.  Understanding how the initial system will grow seamlessly without a complete re-write of the code or a forklift upgrade of the hardware only comes from having experienced and triumphed over these issues in the past.

MAINTAINABILITY

All the hard work and planning that goes into making a system reliable and scalable can be wasted if system designers forget one simple fact: human beings maintain and administer computer systems.  Factoring in the human element to a system design, including the fact that people always make mistakes, is the final critical element to delivering production systems.  Developing strong procedures that allow for the operational realities of running the platform, including third party software component upgrades (such as OS and database), is often the difference between a reliable platform and one synonymous with downtime.     

EXPERIENCE

It is fine to discuss reliability, scalability and maintainability as the keys to delivering production systems, but there is no substitute for practical experience with real systems.  The fnSys principals founded two successful software start-ups that delivered on the above philosophies.

Commerce.TV was a revolutionary interactive television, advanced advertising and T-commerce service provider.  The company built out an end to end system comprised of a world class data center as well as system deployments with three separate MSOs (cable companies) serving well over 100,000 subscribers.  

Open Development Corporation was a software and services company providing enhanced services platforms to RBOC and wireless customers in the mid to late 1990s.  Open Development's openMedia platform was deployed in 25 sites with 10 separate Telco customers, serving well over 10 million subscribers with a 99.999% uptime requirement. 

The challenges of these major deployments taught us the importance of planning for success up front with system architecture.