| Instance Type | CodeBuild Price (hr) | EC2 Price (hr) |
|---|---|---|
| 7-8G 4vcpu x86 | $0.01*60= $0.60 | $0.17 |
| 15-16G 8vcpu x86 | $0.02*60= $1.20 | $0.34 |
| 16G 8vcpu arm64 | $0.015*60= $0.90 | $0.308 |
| 144G 74vcpu x86 | $0.20*60= $12.0 | $3.06 |
Ignoring image storage, you save 3-4x by provisioning your own EC2 Batch instead of using CodeBuild.
$0.00756 per hour is the cost of AWS Lambda at 128M. That’s 22x less than the 8G 4vcpu EC2.
S3 is $0.005 per 1000 writes, $0.0004 per 1000 reads. No bandwidth cost for in-region transfers -1 kb vs 1 G is the same price. Storage is $0.023 per GB. Huge savings chunking small files and using compression.
For example the zstd git repo has 497 files and is 71224 KB, 69544 KB as a tarball, 56608 KB at zstd level 19. You save 99% of the access charges and 25% of the space.