Companies purchase Amazon AWS Reserved Instances as a way to cut cost on their Instances. When you first purchase blocks of reserved instances, you must pick an Instance Type, Network Type & Availability Zone.
As long as you never change your running Instances, you will save money. If you change your instances, you may end up paying double. You’ll pay on-demand pricing for the new Instances AND you will still pay for the Reserved Instances you purchased.
Over time, as you terminate & re-launch Instances, your Reserved Instances may no longer match your running Instances. We’ll need to reconfigure your Reserved Instances so they match your new configuration so you aren’t paying double.
Getting this just right across all your accounts & machines is tedious. Luckily computers are good at the tedious stuff. RIFactor will (read-only) probe your Amazon account for EC2 Instance & Reserved Instances information. It then does three transformations of the Manifest in order:
- RIFactor matches running Instances with the current Reserved
Instances.
- Input
- Instances
- Reserved Instances
- Output
- Instances (unmatched)
- Reserved Instances (unmatched)
- Used Reserved Instances (partial or full)
- Input
- RIFactor will then combine empty Reserved Instances together.
- Input
- Reserved Instances (unmatched)
- Output
- Reserved Instances (unmatched)
- Combined Reserved Instances
- Input
- RIFactor will then split your Reserved Instances and Used
Reserved Instances up to fit remaining Instances.
- Input
- Instances (unmatched)
- Reserved Instances (unmatched)
- Used Reserved Instances (partially used but not to capacity)
- Output
- Instances (unmatched)
- Reserved Instances (unmatched)
- Used Reserved Instances (partial or full)
- Split Reserved Instances (partial or full)
- Split Used Reserved Instances (partial or full)
- Input
Once RIFactor has probed Amazon for data & ran the Manifest through these transformations, it can show you the plan and execute.
NOTE: The purpose of combining Reserved Instances back together again is to optimize applicability. We want, where possible, the Reserved Instances to be in as big as they can be so we can apply them to the most Instances we can. This counteracts our splitting them up again (to fit new node sizes.)
NOTE: RIFactor considers all of your accounts listed in the config file to be linked together on Amazon’s billing. If you have related accounts such as a billing account (with the Reserved Instances in it) and other working accounts with the Instances running in them, the app will consider all of them as a whole & try to apply the best Reserved Instance reconfiguration. If you have many unrelated accounts, then just create a config file for each account.
ALPHA
This is alpha software. It will not damage any running EC2 Instances on your Amazon account. It will only modify your Reserved Instances. Modifying Reserved Instances does not terminate (or alter in any way) your running EC2 Instances.
You should run the application with –dry-run to see what it would attempt to do before actually running it without.
In all cases you should monitor your account(s) to make sure your Reserved Instances match your running instances during our alpha & beta periods.
Hosted on DockerHub.
Requires that libgmp (apt) is installed. We will create a PPA with packages as time permits.
Requires that libgmp (Homebrew) is installed.
Requires that minghc is installed. Later we could trim this down to just libgmp. I’m not a Windows expert though. Pull requests to the Build section of this document are appreciated.
rifactor --helpCreate a new IAM User on each of your AWS accounts. This user account will be used to access your amazon account & modify Reserved Instances.
Save the credentials (access key & secret key) given to you when you create the new user. The name of the IAM User does not matter. We only need the keys.
Now add a User Policy to your IAM User that allows describing EC2 resources & modifying EC2 Reserved Instances.
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": "ec2:Describe*",
"Resource": "*"
},
{
"Effect": "Allow",
"Action": "ec2:ModifyReservedInstances",
"Resource": "*"
}
]
}On your local filesystem. Create a JSON file with the details of your accounts on AWS. Place the IAM access key & secret key from each user into the config file.
{
"accounts": [
{
"access_key": "<<AWS_ACCESS_KEY_ID_HERE>>",
"secret_key": "<<AWS_SECRET_ACCESS_KEY_HERE>>",
"name": "dev"
},
{
"access_key": "<<AWS_ACCESS_KEY_ID_HERE>>",
"secret_key": "<<AWS_SECRET_ACCESS_KEY_HERE>>",
"name": "qa"
},
{
"access_key": "<<AWS_ACCESS_KEY_ID_HERE>>",
"secret_key": "<<AWS_SECRET_ACCESS_KEY_HERE>>",
"name": "stage"
},
{
"access_key": "<<AWS_ACCESS_KEY_ID_HERE>>",
"secret_key": "<<AWS_SECRET_ACCESS_KEY_HERE>>",
"name": "prod"
}
],
"regions": [
"NorthCalifornia",
"NorthVirginia",
"Oregon"
]
}The exact format of “regions” is located here. I’ve also listed it below. This should only change when Amazon adds new regions.
| Region | Location |
|---|---|
| Ireland | Europe / eu-west-1 |
| Frankfurt | Europe / eu-central-1 |
| Tokyo | Asia Pacific / ap-northeast-1 |
| Singapore | Asia Pacific / ap-southeast-1 |
| Sydney | Asia Pacific / ap-southeast-2 |
| Beijing | China / cn-north-1 |
| NorthVirginia | US / us-east-1 |
| NorthCalifornia | US / us-west-1 |
| Oregon | US / us-west-2 |
| GovCloud | AWS GovCloud / us-gov-west-1 |
| GovCloudFIPS | AWS GovCloud (FIPS 140-2) S3 Only / fips-us-gov-west-1 |
| SaoPaulo | South America / sa-east-1 |
You need the GHC 7.8.x compiler & cabal-install (Homebrew, minghc or Ubuntu PPA will work). Review the Travis CI Config File for build steps.
Create a fork & submit a pull request if you would like. Github issues is the place to file your desires and grievances.