Custodian currently provides policy definition around the following AWS resources:
- Amazon Machine Images (AMI)
- Auto Scaling Groups (ASG)
- Elastic Block Store Volumes (EBS Volumes)
- Elastic Cloud Compute (EC2)
- Elastic Load Balancers (ELB)
- Key Management Service (KMS)
- Relational Database Service (RDS)
- Simple Storage Service (S3)
Custodian uses a flexible query language for filtering resources to a particular subset that allows for compound querying. This essentially allows you to filter for things like instances with EBS volumes that are not set to delete on instance termination or stopped instances. This filtering can take into account external data sources. It also provides for resource specific actions around deletion, stopping, starting, encryption, tagging, etc.
The stateless design of Custodian greatly simplifies feature development and operations. It also provides flexibility around execution environment (local cli, server, lambda).
When a user runs Custodian, Custodian will run the specified policy against the account and region specified by the user. Custodian will iterate over all resources defined in the policy. In the CLI, users specify the account and region they want to target. During the run, each policy in the config will generate metrics that are sent to CloudWatch in the account that is targeted. The run will also generate structured record output and logs that will be sent to an S3 Bucket and CloudWatch Log Groups in the account Custodian was run from. If Custodian is being run without Assume Roles, all output will be put into the same account. Custodian is built with the ability to be run from different accounts and leverage STS Role Assumption for cross-account access. Users can leverage the metrics that are being generated after each run by creating Custodian Dashboards in CloudWatch.