One of the things that attract start ups and enterprise customers to cloud deployments is on-demand pricing: you only pay for what you use. This is certainly the cornerstone of AWS billing principles. In fact, some AWS services, like VPC, Elastic Beanstalk, CloudFormation, Opsworks, and IAM are available at no charge at all. You will only be billed for the compute resources you actually run.
Nevertheless, accurately anticipating your monthly costs can sometimes be tricky, so here are five quick tips for managing the Billing on your AWS account.
1. Pay Less when you reserve AWS instances
If you’re currently using an on-demand EC2 or RDS instance, take a few minutes to do a review. If it turns out that your usage is fairly steady over a long period, you may find that reserve pricing can cost you quite a lot less. One up-front payment covering a period of between one and three years can go for a significant discount over the hourly usage for the same instance.
Comparison between On-Demand and Reserve pricing for periods of one and three years:
2. Bid on the Spot market for unused AWS Resources
When you launch an instance, you’re given the option of bidding for unused EC2 capacity on the Spot market. When your bid price is equal to or higher than the current Spot price, you will automatically be given use of an EC2 instance of the specified type for as long as the price doesn’t rise above your bid. Spot bidding can save you up to 90% on the cost of Ec2 instances.
Spot instances are useful for heavy lift tasks such as complex analytics, Big Data processing, scientific computing, and media processing. You just have to be confident that your application can survive unexpected terminations should the Spot price rise above your bid. It may be necessary for you to review the Spot price history in your availability zone.
3. Set CloudWatch alerts to monitor your AWS usage
Use Amazon CloudWatch to have alert notifications sent to you when your monthly charges for using AWS product reach a pre-configured threshold. This frees you from having to log in AWS Billing Console to check for yourself. More importantly, it protects you from the surprises that you can face if you never get around to checking.
How does it work? Suppose your AWS account gets hacked and the intruders decide to spin up few d2.8xlarge instances. When your usage crosses the pre-set threshold, you will receive a message from CloudWatch advising you to take the necessary action.
To get started with CloudWatch, log into the AWS Billing Dashboard, go to Preferences, and select Receive Billing Alerts.
Update: We should also note the huge value of enabling “Receive Billing Reports” alongside “Receive Billing Alerts” – the comprehensive CSV-formatted reports can be a really useful source of valuable insight (hat tip).
4. Tag your resources
When you tag your AWS resources to identify their function or association (QA, Dev, Prod, etc) you’ll be able to quickly know which environments are incurring the highest costs and which business unit is exceeding their AWS cost allocations. This can be a particularly effective way to monitor – and control – costs.
5. Analyze AWS billing reports
AWS provides monthly billing reports that highlight the costs incurred by individual AWS services, and the number of hours used in a month. But some times this isn’t enough when, for instance, you also want to see a forecast of your future expenses or when you’d like to share figures with a customer.
To get this kind of clarity, you can export your billing reports in CSV format or even build a custom application with its own analytics. You could also enable programmatic access to your AWS account and specify an S3 bucket into which you want your billing data copied. AWS will then generate estimated monthly bills several times a day and save them to your bucket. You can also use the S3 bucket data as input for your application. 
