What are Spot Instance Interruptions?
Spot Instance interruptions occurs when the Spot price is greater than your bid price, or when Amazon reclaims Spot Instances to meet higher-priced Spot requests from other customers. These disruptions can happen unexpectedly, typically offering you with a two-minute notice before the instance gets taken back.
Your Spot Instance is terminated and any workloads which are running on it are halted on an interruption. As such, it is critical to build applications which can gracefully take in disruptions by preserving work in progress and reducing disruption or data loss through the utilization of distributed computing frameworks or checkpoints.
Amazon Web Services – Introduction to EC2 Spot InstancesWhat are AWS spot instances?
In this article, we will look into the process of profiling your applications and workloads to confirm if they can take advantage of Amazon EC2 spot instances. Spot instances are spare compute capacity available to you at steep discounts (often as much as 90% cheaper) compared to on-demand prices. The reason you need to qualify for the spot is that the spot is unused and easy to fill, and spot instances can be interrupted by EC2 with a two-minute notification when EC2 needs that capacity back.
What are AWS spot instances?
The Amazon EC2 Spot Instances help you take advantage of the unused EC2 capacity in the AWS cloud. So, Spot Instances provides up to 90% discount compared to on-demand prices. You can simply use the Spot Instances for many features, like fault-tolerant, stateless, or flexible applications such as CI/CD (Continuous Integration/ Continuous Integration/Continuous Deployment), big data, some containerized workloads, web servers, and development workloads. Because the Spot Instances are tightly integrated with some of the AWS services like CloudFormation, Auto Scaling, Amazon Elastic MapReduce(EMR), Amazon Elastic Container Service( ECS), AWS Batch, and many more, users can easily choose how to launch and maintain the applications that are running on Spot Instances.
Users can easily combine the usage of spot instances with on-demand, RI, and savings plan instances. So, the users can further optimize the workload with the help of performance. Also, due to the operating scaling of AWS, Spot Instances can also offer high-level scaling and cost-saving ways to run hyper-scale workloads. There is also the option to hibernate, stop, or terminate your Spot Instances of EC2 when your EC2 reclaims the capacity back into a notice of two minutes. But this will only happen on AWS, where users have easy access to unused compute capacity on such a massive scale.
So while this doesn’t happen all that frequently you do need to be able to handle interruption in order to use the spot effectively. But before moving ahead, we should keep in mind the good practices of the cloud such as:
- Scalability
- Disposable resources
- Automation
- Loosely coupled applications
- No single point of failure
If you’re in a great position to start using spot instances today but instead of diving into all of these cloud best practices let’s just answer one question “What would happen if a server was taken away?”
If your job is time critical the instance must be automatically replaced without human intervention and it should be transparent to your users. That means if it’s big data batch processing or queue processing, the job should automatically restart on a different instance as soon as possible.
If it’s a website your users should be none the wiser. The load balancer automatically sends them to a different instance.
If you aren’t architected for the cloud yet but have a time-insensitive workload you want to save up to 90% on well you’re in luck you can ask AWS to hibernate the instance. When AWS needs the capacity back and will hibernate your memory to disk and then it will start it up again when spot capacity is available just like shutting and reopening your laptop.
Contact Us