Skip to main content

Is your cloud deployment much more expensive than it should be?

closeup photo of 100 US dollar banknotes
If you take any of the cloud platforms, you can spend days looking through all of the various features. AWS alone has over 212 core services. In recent years, a lot of those new features have been created to make it easier to deploy projects into the cloud, and there's nothing wrong with that. However, you have to realize that by using these easier to use services, the cost you end up paying will be much higher than it could be.

The problem with these one-click deployment systems is that they have to assume a lot of things. They deploy infrastructures in your name, including Windows or Linux instances, load balancers, DNS configuration, networking, and so on. While you can manually go in and tweak these resources, if you've relegated the deployment to Amazon, you might not want to then go in and start tweaking the result.

Yet there are many ways that deployments can be improved, both in efficiency and cost optimization, if you have somebody with deep knowledge of the offerings and how they can be tweaked. One such tweak is the concept of reserved instances. If you run your deployment for at least 3 years, you could save 60% off the cost of every instance in use, versus relying on on-demand hardware. You could also resize your instances if you notice the load is lower than you expected, or reduce the minimum number in your auto-scaling group. But that requires proper monitoring.

Another potential cost saving is when it comes to databases. Hosted databases such as RDS are typically one of the bigger cost of any cloud deployment. Again, the size and quantity you need will be fully dependent on the amount of data you have and its utilization. The same is true for static data, outside of databases. If you have a large quantity of static data, such as pictures and videos, you may want to use Amazon S3 instead of EBS volumes, which could save up to 50% of all your storage costs.

The bottom line to all of this is that the cloud has been made easier than ever to use for any purpose, from large corporate deployments to small projects by independent developers. But understanding the principles behind how these features work, and how to optimize them, could make a massive difference in price and performance. These examples are all real world cases I've personally seen in the past years. And this is where hiring a consultant, while it increases costs in the short term, could save massive amounts in the long run.