Amazon Web Services Indiahas launched Spot instances that are a cost-effective option compared to AWS On-Demand instances if the enterprise is flexible pertaining to when the applications run and if the applications can be interrupted. Organizations can also set up Auto Scaling to use Spot instances instead of On-Demand instances. However, before Auto Scaling of web is done with AWS Spot or On-Demand instances, it is important to be abreast with launching and managing Spot instances harnessing Amazon EC2.
Auto Scaling, which comes as AWS support services, helps the organizations ensure that they have the right number of Amazon EC2 instances to handle application volume. There are options to create collections of EC2 instances, also called Auto Scaling groups. The minimum number of instances can be specified for each group, while Auto Scaling ensures that the group does not go below this size. Likewise, the maximum number of instances can also be specified for each group, and Auto Scaling will ensure that the group does not go beyond this size.
So, here’s how enterprises should use Auto Scaling to launch AWS Spot instances or AWS On-Demand instances:
- Create Launch Configurations with a Bid Price – Firstly, it is important to create launch configurations and then set the bid price to use Auto Scaling to launch Spot or On-Demand instances. However, note here that a single launch configuration cannot be used to launch both Spot instances and On-Demand instances.
- Defined-Duration Workload – To ensure that EC2 works better with defined-duration workload, cloud computing service providers helps launch Spot instances that can run continuously for a finite duration, ranging from 1 to 6 hours. Pricing will be based on the requested duration and available capacity. Price of Spot instances for defined-duration workload are generally 30% to 45% less than On-Demand instances.
- Changing the Bid Price – To change the Spot bid price, it is important to create a new launch configuration with a new bid price, and then link it with your Auto Scaling group.
- Maintaining your Spot Instances – When the Spot instance is terminated, Auto Scaling will try to launch a replacement instance to establish the desired capacity for the group. If the bid price is more than the market price, then it will launch Spot instance. Alternatively, it will keep trying.
- High Performance Computing – Organizations executing critical financial analysis or laboratories performing scientific analysis can create clusters that can be formed by thousands of machines.
- Big Data and Analytics – With the aid of spot instances, bulk amount of data can be processed in the most cost effective manner and by harnessing other tools such as Amazon Elastic MapReduce or Hadoop.
- Batch Media Encoding or Analysis – This is another instance of delayed job where cost savings can be achieved using Spot instances and Auto Scaling.
- Testing Scenarios – Testing scenarios like warfare, generic simulations, disaster recovery, etc. are some instances where Auto Scaling can be used to launch spot instances as well as AWS On-Demand instances.