Lambda Self-Managed Code Storage: More Than Just a Quota Bump
Lambda now lets you point to your own S3 buckets for deployment packages, sidestepping the 75GB limit and speeding up deployments.
Editorial summary and commentary based on the original from AWS What's New. Read the original
Lambda is now less opinionated about where your code lives, and that's a good thing.
What changed
- Lambda supports referencing code directly from customer-managed Amazon S3 buckets.
- The default Lambda-managed code storage quota per Region per account increased from 75GB to 300GB.
- No additional Lambda charges apply; customers pay standard S3 storage and data transfer costs.
Why it matters
This change directly addresses a long-standing operational friction point for teams deploying a large number of Lambda functions or significant code sizes. Previously, exceeding the 75GB Lambda-managed storage limit required support tickets for quota increases. By allowing Lambda to reference code directly from an S3 bucket, teams can now manage virtually unlimited function and layer code sizes, dictated only by S3 bucket limits. In practice: this means faster function updates and deployments for those hitting the old limits, as the internal Lambda copy step is removed. It also centralizes code artifact management within a customer's own account, aligning with a single source of truth principle.
The catch
While the announcement frames this as a quota increase and a new storage option, the honest version is that this shifts operational responsibility. Customers must now manage S3 bucket permissions (s3:GetObject, s3:GetObjectVersion) and ensure the Lambda service principal has read access. Furthermore, while Lambda itself incurs no extra charge for this storage mode, customers will incur standard S3 costs, and potentially cross-Region data transfer costs if the S3 bucket is not in the same region as the Lambda function. This also means the 300GB default quota is now less relevant for those who opt into self-managed storage.
Ship it
If your Lambda functions or layers are approaching or exceeding the 75GB (or now 300GB) Lambda-managed storage limit, or if you prefer to manage code artifacts directly in S3, configure your deployments to use S3ObjectStorageMode: REFERENCE. This is particularly relevant for CI/CD pipelines using AWS CLI, CloudFormation, or AWS SAM, where the S3ObjectStorageMode parameter can be set during function or layer creation/update. Pairs with: Amazon S3.
Bottom line: Lambda now lets you use your own S3 buckets for code storage, removing the 75GB limit and speeding up deployments at the cost of managing S3 permissions and costs.
— Filed to /aws
Source (AWS What's New): AWS Lambda announces self-managed code storage