😴 Idle Resources
Idle cloud resources are like leaving a light switch on in a room after everyone has left.
If nobody is using the cloud resource that you are paying for running, it is eroding the business value of your cloud investment, but uncovering those idle or zombie resources can be time consuming and requires knowing where to look.
Vega Cloud simplifies this process by automatically surfacing opportunities in your cloud portfolio and presenting them to you to disposition.
When would an "Idle Resource" not be considered waste?
There are cases where an Idle Resource may be part of an ongoing development/testing cycle. The most reliable way to validate if this is the case is to talk to the team responsible for the resource to layer in the _business context_ around the recommendation. If the resource is allocated as "production" however, then the prior case is less likely.
Do I have any Idle Databases?
Database instances that have been billed for at least 90% of the last 30 days without having received any connection will be flagged as idle.
Do I have any Idle Load Balancers?
Load balancers running for 90% of the last 30 days, but are not transferring any data in or out, are flagged as idle.
Do I have any Idle NAT Gateways?
Any NatGateway running for 90% of the last 30 days (about 4 and a half weeks), but is not transferring any data in or out, is flagged as idle.
Do I have any Idle VMs?
Any virtual machine (VM) resource type with a max CPU utilization under 5% for the last 30 days (about 4 and a half weeks) is flagged as idle. These workloads should be evaluated for terminating, or migrating/consolidating to shared tenancy.
How do you know the max CPU utilization for the recommendation?
Vega will leverage CPU metrics from your cloud native monitoring source (e.g. Azure monitor, AWS CloudWatch, etc), or an alternative metrics source such as DataDog. Alternative metric sources can be integrated upon request.
Not every workload will be CPU bound, does this recommendation provide false positive?
It is true, some workloads might be more network, GPU, memory, or storage bound. Vega chooses to leverage CPU max utilization over a 30 day window as the proxy measure as it is an efficient and reliable screener, especially when combined with other dimensions such as environment tier. If the CPU is low utilization AND it is allocated as DEV, then this increases the confidence that waste has been identified.