The COST of Abandoning Infrastructure


Recently our Innovation Studio partnered with a startup to build a flagship SaaS application they would sell as a service to their clients. Like most startups cash flow was a big concern. They needed a robust solution that could handle the potential volume without constantly re-architecting, but they needed it on a budget. Whether you’re a startup, established business or a large enterprise, getting the most for your dollar and avoiding idle expenses is always a priority.

The assessment for this effort was fairly quick thanks to our Innovation Studio’s passion for staying up to date on the latest releases and capabilities of the cloud.  We were already familiar with a few patterns to draw from and chose AWS’s many PaaS offerings to architect out a robust, scalable, pay-as-you-go solution.

We were able to develop the basis of a new application in hours vs. days by leveraging AWS to setup a HTTP enabled S3 bucket along with the power of Lambda services triggered through API gateway manipulating data on RDS database instances. There’s no waiting on expensive complex infrastructure here and our client is loving the fact they aren’t paying for IT resources to manage servers, things scale with use and they only pay for the compute power and data they actually use.

This design has quickly become one our most used patterns and again and again both our clients and engineers are impressed with the flexibility, scalability and cost for building great online applications. Pair these items with Cognito security, Cloud Front regional caching and the many other offerings within AWS and there’s no limit to having secure, high performing, feature rich solutions in less time and for less money.

Understanding the cost saving can be complex so I’ll do my best to break this down visually. For our scenario we’ll assume you have a API driven application that does 1 million transactions a day averaging 3s execution time using 128mb of memory (a lot for basic individual functions, database updates, reads, etc.) and your users are corporate so they’re only really active 8am – 5pm daily. For the EC2 instance we’ll assume the cheaper of the general use instance M3. Take a look at the cost structure breakdown: 

As you can see there is quite a difference in cost running an API layer on an EC2 instance vs. Lambda, not to mention terrible efficiency with so many unused hours burning cost to your bottom line. I know, I know, “but you can do more on an instance!”, we’re aware you can also host your Web app and DB on that instance doing those calculations along with Lambda leveraging S3 and RDS will still yield you considerable savings a month. Let’s not forget the intangible here. With that Ec2 instance you’re responsible to ensure it’s patched, up to date, running, can handle the load of a growing user base, etc., while the Lambda platform will just scale as needed making the most of its natural elasticity.

While we understand this solution isn’t always the right one for every situation, you can’t deny the potential cost savings leveraging a cloud solution can bring to a number of scenarios. Our Innovation Studio focuses on building solutions with the latest and greatest technology, we’ve dug into it so we’re ready to help when you need us most.

About The Author

Andrew Duncan is a Director of Software Engineering in Richmond, VA. He is a driven technologist focused on modern technology stacks and best practices. Andrew believes nothing is more rewarding than making software needs a reality with a focus on flexible, scalable and supportable code.