Cloud services, like those from Amazon AWS and Microsoft Azure, make managing and deploying your infrastructure more efficient, and help improve quality. Before the cloud (BTC), establishing an infrastructure required you to:
- Buy servers and network equipment
- Configure servers and network equipment
- Build or lease space in a data center that would provide ping (network connection) and power
- Build and configure the Firewall, load balancer, servers, and any other parts of your infrastructure that you needed
- Install the software needed for your solution (for example, Microsoft Internet Information Services (IIS) and the software used to run your product)
After this was all setup and configured, you would still need to manage and oversee everything on a continual basis, looking for hardware failures, software issues, security attacks, and other obstructions.
The way infrastructure was built provoked what psychologist, Carol Dweck, calls a fixed mindset. Infrastructure was time consuming to setup, capital intensive, and hard to change. You would have to plan your capacity carefully, because expanding it could take weeks. This means that typically, you would need to plan for the worst case load, and design your infrastructure around that scenario. This process works, but it also adds an exponential amount of waste; most of your infrastructure would only be used a small fraction of the time, and would then sit idle for the rest.
The cloud flips all of this, and therefore requires a change in thinking. Instead of being seen as expensive hardware that is difficult to install and manage, infrastructure becomes like software: it is something you can spin up and down on demand. It moves from a fixed asset to one that is variable, based on need, one that can be delivered on demand. The cloud also changes the payment model to match this new mindset: you pay by the hour instead of with large, upfront capital costs. Once your mindset shifts, there is a variety of interesting things that you can do with the cloud, such as:
- When you deploy a change, instead of installing it on production, you simply spin up a whole new server, deploy the change, test, and once verified, simply switch servers and dispose of the old one.
- You can have development servers spun down during evenings and weekends, saving you money when they are not used.
- You can setup auto-scaling to have one server responding to load, and then multiple other servers can be spun up when demand is high and turned off when demand goes down. (Since most sites have spikes in demand, auto-scaling can turn into significant savings.)
- You can script your entire infrastructure, then store it in version control and run it to spin up duplicates for short periods of time. For example, if you need to perform a load test on your production infrastructure, you can duplicate it for the rest of the test and then throw it away.
Of course, this change in mindset and method isn’t easy. Even today, many enterprise software companies haven’t adapted to it. They are still pricing licenses by CPU, which means that we have to limit our infrastructure not by technology, but by license. This complacency in infrastructure management prevents companies from capitalizing on one of the main drivers of progress, improvement. Those companies are still stuck in the fixed mindset and haven’t adapted to competition in the ever-evolving digital world. Some companies provide cloud offerings themselves, which works if you only want their product, but it doesn’t always work well if you are integrating the product with other systems.
Now, more companies are seeing the value in subscription models based on usage. The Co-CEO of Autodesk explained in a CNBC interview how this model not only attracted more customers through dramatic reductions in up front costs, “but what keeps them coming, is the value we start to deliver.” Easier access to the software, new types of automations, and new types of process management tools, lead to 233,000 soft subscriptions to Autodesk’s cloud-based service in the first quarter.
Mindsets are difficult to change, and it generally takes a while for the biggest benefits of that change to show, but once the change occurs successfully, whole new possibilities open up. The cloud mindset allows you to look at infrastructure in a completely different way than how you have in the past: it allows you to see your infrastructure from your user’s point of view, and look beyond metrics like uptime and incident rates to focus on the end user experience.
Carol S. Dweck Ph.D. (2007, December 26). Mindset: The New Psychology of Success. https://www.amazon.com/Mindset-Psychology-Success-Carol-Dweck/dp/0345472322
Fred Beer (2017, July 11). 3 Drivers of Progress.
Kevin Dolohanty (2017, June 19). J. Crew Vs. Domino’s: Competition in the Digital Age.
CNBC (2017, May 19). Autodesk Co-CEO: Subscription Model is Changing the Cloud. http://www.cnbc.com/video/2017/05/19/autodesk-co-ceo-subscription-model-is-changing-the-cloud-game.html