According to Gartner, 2019 will see investment in the public cloud market to reach $206.2 billion. Over time, as cloud usage progresses from niche to broad deployments, most enterprises usually discover the critical importance of cloud cost management after receiving few bills, and in some cases ‘bill shocks’.
Rightscale conducted a study on how much of the cloud providers’ revenue comes at the expense of cloud users not understanding how they can prevent wasted cloud costs. According to Rightscale report, cloud users estimate that they are wasting 40 percent of their cloud spend. Segment.com published their story of saving $1 million (a must read article) in their cloud bill by identifying cost wastage opportunities. Their experience tells us how the ability to provision thousands of dollars worth of infrastructure with a single API call can waste millions if cost management not done correctly. Just to give some numbers, between idle resources and over provisioning itself, wasted cloud spend will exceed $14.1 billion in 2019.
In this 3 part article series, let us have a look at few of the key cloud wastage examples for major public cloud platforms — AWS, Azure and GCP. AWS being the most widely used cloud platform, Part 1 article looks at the cost wastage examples for AWS. Feel free to use below listed opportunities for a quick start on your cloud cost savings:
- Running ec2 instances from a lift and shift VM migration that are running 24x7x365 and not having any reservations purchased for them.
- Spinning up resources that aren’t used and charged by the hour (ec2 instances, load balancers, VPC endpoints, etc)
- Not properly cleaning up all resources when deleting resources — eg Orphaned snapshots. EBS volumes get deleted once EC2 instances are terminated, but snapshots are left to rot.
- Wrong architectural decision thinking application needs IOPS and opting for provisioned IOPS, when general purpose would do the trick
- Using CNAMEs instead of ALIAS records for route 53 records pointing to AWS endpoints
- S3 Buckets which receive no requests yet sit on the standard tier.
- S3 buckets with versioning enabled and no version expiration configured