Configuring EC2 Auto Scaling for Performance & Efficiency

AWS EC2 Auto Scaling provides true elasticity, flexing to meet application demands and powering many cloud applications that have variable workload patterns. However, to make the most of AWS Auto Scaling, cloud architects and engineers must leverage machine learning to select the right EC2 instance types, sizes, and group scaling parameters.

Densify machine learning technology analyzes workloads across all Auto Scaling nodes and groups, and intelligently recommend the best node types and sizes, as well as scaling parameters (min and max group sizes). This helps cloud engineers better align the entire ASG with the actual demands of the application, maximizing app performance and resource efficiency.

Intelligent EC2 Auto Scaling Management Demo

Ready to see the right configurations for your EC2 workloads?

Request a 1:1 Demo

Complete Video Transcript

I'm going to show you how you can use Densify to intelligently manage your AWS Auto Scaling groups and ensure they have the precise resources to enable your applications to perform optimally.

When it comes to configuring AWS Auto Scaling groups, cloud engineers have a massive challenge ahead of them. First they must identify exactly the correct instance type suited for the application running on that Auto Scaling group. Then they have to worry about the minimum and maximum size nodes in order for the application to perform effectively.

The good news is that Densify intelligent analytics and our Cloud-Learning Optimization Engine make this task much easier and achievable.

Here's an example—an AWS Auto Scaling group that is running in the US West 2 region for the last 60 days. We have actually configure this Auto Scaling group to think that it is a computer-optimized workload, and we placed it on a C4 extra large Linux box.

Now, Densify analytics have looked at the cloud, watched the resource metrics, and studied the CPU, memory, disk, and network I/O intelligently to tell you what is the precise Auto Scaling group configuration that you need.

Let's take a look at this and see how this is happening.

At the bottom of the screen, you're looking at the CPU utilization. And, in this case, over the last 60 days using our machine learning and predictive analytics. We're looking at the CPU utilization for this particular AWS ASG, and as you can see, this thing is averaging just around 9% CPU utilization, and it peaks slightly throughout the day, towards the end right at 40%, but it comes down.

The next thing that we say is, hey, what if we could also analyze the busiest instance that is part of that Auto Scaling group, and see what exactly that is also doing on the busiest state.

And in this case, the business member average is just around 17% on the busiest state, so not really a computer-optimized workload, because you would expect the CPU utilization to be much higher.

Now, we also investigate the memory utilization, and in this case, you can see the memory stays relatively stable throughout the day at around 50%, and throughout the day, network I/O and the disk usage fluctuate slightly.

So, what Densify does is we take all this resource utilization data, the CPU, memory, disk and network I/O, and then we evaluate this against the entire catalog of ASG options to determine the best match.

And we ask, hey, what is the best HG configuration that were there that we could suggest here that will give us the best performance at the lowest possible cost. And in this case, the answer is the R5 Large, so we're recommending that, hey, you should be moving this Auto Scaling group from the current configuration that is a C4 XL to our R5 L and we've done this by studying the actual workload and matching it to what it would look like on the R5 Large. And, if you were to take our recommendation, then here's your current ASG configurations that you would need, and you need to change the minimum size from three to one, and then from three to four. And in doing so, you could also see that the performance will improve and the cost will dramatically come down as well.

So, rather than spending $145 per month on it, you'd be spending around $92 per month, on average.

Now, using Densify analytics, we can help you tune, optimize, and manage your Auto Scaling groups more effectively and intelligently, making sure that your Auto Scaling group applications precisely have the right resources to perform optimally.