16
Cost Management Week: Wastage
Hyperscalers provide many mechanisms to optimise spend on Cloud. One of the pieces most people miss when getting started however, is managing cost through wastage. Wastage of resources - whether it's oversized resource choices, redundant and unnecessary backups, or left over deployments from automation - we often stumble upon large amounts of spend that didn't need to happen at all.
Each hyperscaler has a different method to provide cost insights. For Azure, you can use the Cost Analysis blade, which gives you insight by resource group, resources deployed, or service type. You can get a simple breakdown at a macro level showing the type of spend, and monitor this for areas that exceed what you expected. For instance, the author noticed that 24 Days of Cloud Ready itself wasn't being cached, leading to greater-than-expected traffic hitting the origin. This is quickly mitigated, and the cost analysis reports are updated quickly.
To make it easier, hyperscalers also provide Advisor Recommendations - where they programmatically determine cost recommendations - including (but not limited to):
Another method - which may surprise you - is to ask your hyperscaler's representative. Every Cloud provider in the world hopes you use their service for longer, and would rather you were able to spend $10/mo for a long period, than $1000/mo once, be unhappy, and turn the service off. Through engagements like Well Architected Cost Reviews, your hyperscaler will be able to help you on this journey.
Finally, use Cost Alerts - which analyse your baseline spend against pre-define targets to ensure that any unexpected cost (such as a large instance that was accidentally deployed) can be discovered and mitigated before that cost is incurred. Cost Alerts can be configured to monitor both against a static budget, or against anomalies, to ensure that you're able to hit your Cloud Ready goals.
Each hyperscaler has a different method to provide cost insights. For Azure, you can use the Cost Analysis blade, which gives you insight by resource group, resources deployed, or service type. You can get a simple breakdown at a macro level showing the type of spend, and monitor this for areas that exceed what you expected. For instance, the author noticed that 24 Days of Cloud Ready itself wasn't being cached, leading to greater-than-expected traffic hitting the origin. This is quickly mitigated, and the cost analysis reports are updated quickly.
To make it easier, hyperscalers also provide Advisor Recommendations - where they programmatically determine cost recommendations - including (but not limited to):
- Oversized VM Instances
- Unused Storage Accounts
- Reserved Instance Opportunities
Another method - which may surprise you - is to ask your hyperscaler's representative. Every Cloud provider in the world hopes you use their service for longer, and would rather you were able to spend $10/mo for a long period, than $1000/mo once, be unhappy, and turn the service off. Through engagements like Well Architected Cost Reviews, your hyperscaler will be able to help you on this journey.
Finally, use Cost Alerts - which analyse your baseline spend against pre-define targets to ensure that any unexpected cost (such as a large instance that was accidentally deployed) can be discovered and mitigated before that cost is incurred. Cost Alerts can be configured to monitor both against a static budget, or against anomalies, to ensure that you're able to hit your Cloud Ready goals.