One of the top reasons companies are moving to AWS is the ease of scalability. Companies running news sites, promotional campaigns etc. can truly benefit from this ability. The challenge is that AWS instances don’t come pre-configured to handle traffic spikes right out-of-the-box. Let’s take a look at how a static environment vs. one set up with auto scaling handles traffic spikes.
In this example, the environment contains an elastic load balancer (ELB) along with three elastic computing cloud (EC2) instances. If left open to the internet during a high volume time period this AWS user would only be able to scale up to whatever traffic 3 instances can handle. TWhile this may be acceptable in some cases, often times it may result in downtime. In order to address this challenge an AWS user should introduce an Auto Scaling Group (ASG) to their infrastructure.
An ASG (s a policy applied to the load balancer that sets a certain threshold based on Network IO, CPU or memory of the instances), dictates the scaling. Once hit, it will spin off additional EC2 instances to power the site. There is also a reverse metric in the policy that says that once traffic is reduced back to normal levels (based on the metric you set in advance) the incremental EC2 instances will be removed so you are not paying unnecessary fees.
Common ASG settings include:
- Standard number of instances– number of EC2 instances during normal traffic levels
- Max number of instances– total number of instances the infrastructure can be scaled to
- Minimum number of instances– there may be instances where a lower number of instances than the standard configuration are used
As part of Connectria’s Managed AWS services, our engineering team goes in and takes care of all the settings on the backend, so your site is ready to handle unexpected (or expected) traffic spikes without any glitches.
In the next write-up on auto-scaling, we will provide more detail on ways to design your application to utilize auto scaling effectively. If you would like to learn more about auto scaling or about Connectria’s AWS solutions please shoot us a note.