CODEWITHSUNDEEP
amazon-web-servicescloudautoscaling
Ryan Besari
Ryan Besari

Reputation: 31

Forecasting for autoscaling AWS

it is possible to make forecasting for auto scaling in Amazon Web Server? if it possible how to implement or configure it?

Upvotes: 0

Views: 161

Answers (2)

Mahdi
Mahdi

Reputation: 3349

As John says the forecasting feature does not exist today, but if there is a pattern in the behavior of your auto scaling actions, you can create scheduled actions. For instance, take a look at the following graph:

enter image description here Based on this, I can tell that the desired number of instances increases on Wednesdays to the max (15 here) and decreases significantly on Sundays. There is a logic behind this behavior and I know it. So it might make sense to schedule a scale out action on Wednesdays and a scale in action on Sundays. You can go deeper in your data and find more complicated patterns but I personally, have not seen many use cases for that.

Upvotes: 0

John Rotenstein
John Rotenstein

Reputation: 269400

Autoscaling can react to metrics, such as the CPU Utilization of Amazon EC2 instances, the length of an Amazon SQS queue and application latency measured by Elastic Load Balancing. However, these are all reactionary activities.

There is no ability to forecast load and proactively scale.

Netflix built their own tool called Netflix Scryer:

"Scryer predicts what the needs will be prior to the time of need and provisions the instances based on those predictions."

However, unlike many other tools written by Netflix, it is not available as open-source.

Upvotes: 1

Related Questions