Insight Venture Partners held a software developers forum yesterday in Boston. Several interesting trends emerged from a survey of the attendees. The forum also had several presentations on best practices for software development.
Trends and survey results
Business model - ASP model 42% versus 58% traditional software model. This is a pretty big shift that has been happening over the past 5 years. Salesforce.com is the poster child for the ASP model, or as we call it, SaaS - Software as a Service.
Database usage - SQLServer 64% versus MySQL 36%, interestingly, no Oracle or DB2. Start-ups are more cost sensitive than larger enterprises so it is not too surprising that there were no Oracle or DB2 users.
Development Tools - 28% Microsoft.Net, 22% Java, 13% C/C++, 37% Other
Development Process - 38% Agile Process/eXtreme Programming, 34% RAD/Iterative, 28% traditional waterfall approach. Fast development and multiple releases per year is the norm for start-ups. The newer development processes are more geared to this kind of environment.
Over 80% of attendees used SalesForce.com. I was amazed that so many of these companies used Salesforce.com and not some of the other ASP CRM vendors.
Less than 10% were filing patents for their inventions. Too time consuming and too expensive.
Nearly 100% of attendees were not now, and would not in the future, consider outsourcing development or support off shore. This might be skewed from the normal population since all of these companies were start-ups and usually deep into the development process.
Software Development Process
Extreme programming (XP) and Agile Process are relatively new software development processes that stress breaking the work down into bite size chunks that can be done in 2 to 3 weeks. That is right...2 to 3 week development and test cycles. The idea is to demo completed features every two to three weeks to potential customers to get immediate feedback. That doesn't mean they release and SHIP code, but they do demo the code early and often.
I used the XP model at a start-up a few years ago. It was amazingly successful in terms of meeting dates, getting feedback, keeping engineering morale high, and sustaining momentum. Much better than the traditional "big bang" model where you code for 6 to 9 months and then hopefully deliver what customers want. I will write a separate article on the XP process sometime in the future.
The XP model is great for engineering and product management. Architecture, QA and Documentation groups tend to have a tougher time adapting to the process. They tend to be longer cycle tasks and more "end-to-end" oriented. They have a hard time dealing with discrete, short term projects. About 72% of the companies at the forum were using some sort of Agile Process / XP or RAD development process, and seemed to be successful doing it. This seems to fit with the fact that 42% of them are delivering their software with an ASP model. This server based model tends to support, even demand, rapid updates of fixes and new features. I will write a whole separate story about my experiences with the XP programming model. For now, suffice it to say that I believe it is a great model for delivering software FAST. But, it takes a special breed of manager to make it work effectively long term.
Company statistics
About 65% of the companies were focused on software services, mostly delivered with the ASP model. Surprisingly, about 25% of the companies were delivering software development or productivity tools. I thought this market was saturated, but there were some interesting companies delivering new tools for the AJAX, web based services, ASP types of environments. Security was the focus of 10% of the companies. The rest were all over the place.
So, high level take aways from the developer conference were;
- Software as a Service, or ASP model is growing very fast.
- New development processes are being used to deliver software much faster.
- Companies are being built at far less cost, and much faster, than they were 5 years ago during the dot com bubble. Entrepreneurs are boot strapping their companies, using cheaper technologies, delivering over the web, and waiting longer before taking VC money.
