Get updates by email or on Twitter at @hughdurkin

Remembering Macromedia Coldfusion

My recent post about got me thinking about technology choices earlier in my career. With the benefit of hindsight, I made some pretty bad choices.

One of them was doubling down on Macromedia Coldfusion (now Adobe Coldfusion) in the mid 2000's. My first introduction to it should have given me ample warning.

In 2000 I joined a little startup (within a newspaper group) called Bluepages. Bluepages was a directory and website development service to help small businesses - plumbers, electricians, funeral services, florists - get online. Ultimately, it was way too early for such a small market. Inevitably, it failed.

When I joined, the business directory had been built with Coldfusion, backed up by a Microsoft SQL Server database. It was, if I recall, expensively hosted by BT and fell over on more than one occasion.

We improved over the next couple of years adding news, TV listings, mortgage applications, travel bookings, forums, and other "portal" stuff - but every step of the way our Coldfusion-built directory was a hindrance. It was slow, crashed a lot, and expensive to maintain. We eventually spun up a PHP-built replacement for it.

Fast forward a few years, when I started the company that ran, amongst other things. For whatever reason, I decided to give Coldfusion a second chance. Although some web apps I built with it - like - scaled well, others did not.

Caching, in particular, was a memory-bleeding black box. An expensive one.

I've since apologised to Michele and Paul from Blacknight for the hassle I caused them "back in the day." Manual reboots, RAM replacements, the works - we could never figure out how to make it work. Eventually I cut my losses in favour of the trusty LAMP stack.

After my pivot-bet back to LAMP, I recall scaling a PHP/MySQL price comparison site to 1.2 million unique users per year on 2 x Amazon EC2 small instances (one frontend, one MySQL database). Total cost? Around $70 a month. I had been spending $250 per month per server running Coldfusion. Plus, an annual Coldfusion license. Bonkers.

The big lesson from this? Well, three actually.

  1. Don't let shiny new technology distract you. What's new is what's broken.

  2. Do let shiny new approaches to existing technology enhance what you're building. The Jamstack (and the ecosystem around it) is a new approach - not a new technology. No wonder it's growing as fast as it is.

  3. If Moore's Law isn't working for you, it's working against you. Compute power is cheaper than ever. If you're not making hay on that reality, you're losing.