Capacity Planning for AWS

By supertech01 | Published February 24, 2016

Amazon Web Services provides businesses with an incredible level of scalability for the delivery of applications and services at the click of a mouse. This has revolutionized how companies deliver and manage applications, but without careful planning, it can be easy to under-resource or overspend on AWS. ,  

Since most cloud-based applications need servers from Amazon’s Elastic Compute Cloud (EC2) and managed databases from Amazon’s Relational Database Service (RDS), we’ll focus our attention on those two core services.

Web and Application Servers (EC2)

EC2 instances can consist of Linux or Windows operating system on which your applications can be installed and easily accessed via the web. These instances come in varying levels of horsepower (memory and CPU). It’s important to know how your applications operate when choosing the proper instance size, as the price can grow dramatically if your application has tremendous demands on memory. Choosing the right combination may take some testing, and fortunately provisioning instances (even once your application is installed) is quite easy with EC2’s snapshot capability.

Amazon complements this with several sub-services, including of course storage, network load balancing, firewall security (they call Security Groups), monitoring and even virtual private clouds (to maximize security). Many of these ancillary services bear their own costs, so it’s important to understand the trade-offs.

Cost reduction is good but don’t forget minimum high-availability requirements: the minimal structure for a modern web application is composed by (at least) 2 EC2 instances (deployed in different availability zones) and behind an Elastic Load Balancer.

Database (RDS)

While you may choose to deploy a relational database on EC2, there is a strong case to be made for the built-in management within Amazon’s RDS. The focus is on providing a database while taking the management of the server on which it resides off the table.

Savings on RDS are possible using instance reservations and making adequate architectural choices. Usually databases are the lesser volatile part of infrastructure; choosing a less expensive reserved instance should be easier with RDS than in EC2. While data changes, the RDS Instance could remain the same with different purposes.

Finally, size matters in RDS and is a primary driver of expense. Fortunately, increasing RDS instances storage size can be done with a minimal-near-to-zero downtime.  You can create an instance with a reasonable storage size and keep growing, no need to provision 1TB from the beginning.

With questions about optimizing your Amazon Web Services infrastructure, please contact Superior Technology at 845-735-3555 or online at


Comments are closed.

Archive by Date