Sometimes, your IT systems will need to cater for spikes of demand beyond what you might consider ‘normal’. These spikes are often over short time intervals and are really difficult to plan for.
As a result, businesses often make heavy investments in IT infrastructure well above the height of a potential spike. This is incredibly expensive because it means they’re paying for resource capacity that’s entirely unutilised – with exception to the spike.
This approach is incredibly inefficient, but the alternative is risking the loss of customers or damaging staff productivity.
AWS Auto Scaling dynamically and automatically adjusts your AWS resources to meet demand. As demand increases, your AWS resources increases. When demand subsides, the capacity of your resources return to normal – and so does your bill.
Now you’re having your cake and eating it too. You can meet customer or staff demand and only pay a fraction of the price.
Imagine you own a bakery that makes the best chocolate cupcakes in town. You sell an average of 3 cupcakes per hour, so you bake your cupcakes with a special machine that produces cupcakes at exactly the same rate. Everything is perfect, or at least, everything seems perfect.
It doesn’t take long and word of your amazing cupcakes has spread. Now, during lunch for two hours each day, customer demand has tripled.
To keep pace with demand, instead of 1 special machine, you need to run 3 special machines to produce 9 cupcakes per hour for just two hours each day. The problem is that those special machines cost a lot of money upfront.
If you don’t buy the machines, you’ll send a queue of customers out the door and your cupcakes will have sold out by the time they get inside. This results in a terrible customer experience which impacts your business’ reputation, let alone the loss of sales.
You can’t afford to meet demand. But you can’t afford not to, either.
How much better would it be if your cupcake machines could produce cupcakes at exactly the same rate as customers entering your shop? And what if, rather than paying for each cupcake machine upfront, you paid for those machines on a per-cupcake basis? Now, we’re talking.
AWS Auto Scaling works by keeping watch over a key performance metric for each of your business systems. When the performance metric reaches a threshold, Auto Scaling increases (or decreases) the capacity of the AWS resources that sit behind these systems.
Think of this in the way that a thermostat adjusts the temperature of your home. First, you set the ideal temperature. Then, your air-conditioner heats or cools the house accordingly.
Just as you might have a preferred temperature for the home – with AWS Auto Scaling – you get to determine the best performance metrics that tell Auto Scaling when and how to respond.