Cost management
AWS Deadline Cloud provides budgets and the usage explorer to help you control and visualize costs for your jobs. However, Deadline Cloud uses other AWS services, such as Amazon S3. Costs for those services are not reflected in Deadline Cloud budgets or the usage explorer and are charged separately based on usage. Depending on how you configure Deadline Cloud, you may use the following AWS services, as well as others:
Service | Pricing page |
---|---|
Amazon CloudWatch Logs | Amazon CloudWatch Logs
pricing |
Amazon Elastic Compute Cloud | Amazon Elastic Compute Cloud
pricing |
AWS Key Management Service | AWS Key Management Service
pricing |
AWS PrivateLink | AWS PrivateLink
pricing |
Amazon Simple Storage Service | Amazon Simple Storage Service
pricing |
Amazon Virtual Private Cloud | Amazon Virtual Private Cloud
pricing |
Cost management best practices
Using the following best practices can help you understand and control your costs when using Deadline Cloud and the tradeoffs you can make between cost and efficiency.
Note
The final cost of using Deadline Cloud depends on the interaction between a number of AWS services, the amount of work that you process, and the AWS Region where you run your jobs. The following best practices are guidelines and may not significantly reduce costs.
Best practices for CloudWatch Logs
Deadline Cloud sends worker and task logs to CloudWatch Logs. You are charged to collect, store, and analyze these logs. You can reduce costs by logging only the minimum amount of data required to monitor your tasks.
When you create a queue or fleet, Deadline Cloud creates a CloudWatch Logs log group with the following names:
-
/aws/deadline/
<FARM_ID>
/<FLEET_ID>
-
/aws/deadline/
<FARM_ID>
/<QUEUE_ID>
By default, these logs never expire. You can adjust the retention policy of log groups to remove old logs and help reduce storage costs. You can also export logs to Amazon S3. Amazon S3 storage costs are lower than those for CloudWatch. For more information, see Exporting log data to Amazon S3.
Best practices for Amazon EC2
You can use Amazon EC2 instances for both service-managed and customer-managed fleets. There are three considerations:
-
For service-managed fleets, you can choose to have one or more instances available at all times by setting the minimum worker count for the fleet. When you set the minimum worker count above 0, the fleet always has this many workers running. This can reduce the amount of time that it takes for Deadline Cloud to start processing jobs, however you are charged for the instance's idle time.
-
For service-managed fleets, set a maximum size for the fleet. This limits the number of instances that a fleet can auto scale to. Fleets won't grow past this size even if there are more jobs waiting to be processed.
-
For both service-managed and customer-managed fleets, you can specify the Amazon EC2 instance types in your fleets. Using smaller instances costs less per minute, but may take longer to complete a job. Conversely, a larger instance costs more per minute, but can reduce the time to complete a job. Understanding the demands that your jobs place on an instance can help reduce your costs.
-
When possible, choose Amazon EC2 Spot instances for your fleet. Spot instances are available for a reduced price, but may be interrupted by on-demand requests. On-demand instances are charged by the second and are not interrupted.
Best practices for AWS KMS
By default, Deadline Cloud encrypts you data with an AWS owned key. You are not charged for this key.
You may choose to use a customer managed key to encrypt your data. When you use your own key, you are charged based on how your key is used. If you use an existing key, this will be an incremental cost for the additional use.
Best practices for AWS PrivateLink
You can use AWS PrivateLink to create a connection between your VPC and Deadline Cloud using an interface endpoint. When you create a connection, you can call all of the Deadline Cloud API actions. You are charged per hour for each endpoint that you create. If you use PrivateLink, you must create at least three endpoints, and depending on your configuration, you may need as many as five.
Best practices for Amazon S3
Deadline Cloud uses Amazon S3 to store assets for processing, job attachments, output, and logs. To reduce the costs associated with Amazon S3, reduce the amount of data that you store. Some suggestions:
-
Only store assets that are currently in use or that will be used shortly.
-
Use an S3 Lifecycle configuration to automatically delete unused files from an S3 bucket.
Best practices for Amazon VPC
When you use usage-based licensing for your customer-managed fleet, you create a Deadline Cloud license endpoint, which is a Amazon VPC endpoint created in your account. This endpoint is charged at an hourly rate. To reduce costs, remove the endpoints when you are not using usage-based licenses.