Reserved and Unreserved Concurrency

Managing concurrency in AWS Lambda ensures that high-priority functions always have the capacity they need. By default, your account’s unreserved concurrency sets a soft limit on total parallel executions. For mission-critical workloads, you can allocate reserved concurrency to specific functions.

What Is Concurrency?

Concurrency refers to how many Lambda function executions run in parallel. For example, 100 simultaneous invocations count as 100 concurrent executions.

Viewing Concurrency Settings

To view or modify a function’s concurrency:

Open the AWS Lambda console and select your function.
Click Configuration.
In the left navigation pane, choose Concurrency.

The image shows the AWS Lambda console, specifically the "Concurrency" configuration page, where function concurrency settings and provisioned concurrency configurations can be managed.

Note

We use an SQS cold start demo here. Substitute your own function for real-world testing.

Unreserved Concurrency

Unreserved (account) concurrency is the pool shared by all Lambda functions in an account. AWS sets a default soft limit of 1,000 concurrent executions per region.

All functions combined cannot exceed this limit.
To increase it, submit a support request.

Reserving Concurrency for Critical Functions

Reserved concurrency guarantees a minimum number of concurrent executions for a specific function:

In the Concurrency page, click Edit.
Select Reserve concurrency.
Enter the reservation amount (e.g., 300).
Click Save.

The image shows an AWS Lambda console page for editing concurrency settings, with options to use unreserved account concurrency or reserve concurrency. The "Reserve concurrency" option is selected with a value of 0, and there are "Cancel" and "Save" buttons.

Warning

If your function exceeds its reserved concurrency, additional invocations are throttled with 429 errors.

Reserving 300 concurrency reduces the unreserved pool from 1,000 to 700:

The image illustrates a concept of reserved concurrency, showing a division of 1000 default account concurrency into 700 unreserved and 300 reserved, with 400 invocations depicted.

Use reserved concurrency to prevent downstream overload:

The image illustrates the concept of reserved concurrency in AWS Lambda, showing two scenarios: regulated concurrency and overloaded concurrency, with corresponding icons and labels.

Note

There is no additional cost for reserved concurrency.

Unreserved Concurrency Limits

Your total unreserved concurrency is a per-account limit. In some regions, bursts up to 3,000 concurrent executions may occur temporarily, but sustained usage above the default requires a support request.

The image illustrates a concept of reserved concurrency, showing a division between 700 unreserved and 300 reserved units, with a label indicating "Free."

The image shows the text "Reserved Concurrency" with the number "3000" below it, accompanied by labels "In some Regions" and "Limited Time."

Reserved vs Unreserved Concurrency Comparison

Concurrency Type	Definition	Default Limit	Ideal Use Case
Unreserved (Account)	Shared pool for all Lambda executions	1,000	General workloads
Reserved (Function)	Dedicated capacity for a single high-priority job	N/A	Traffic-sensitive or rate-limited APIs

Summary

Unreserved concurrency sets the total pool for all functions.
Reserved concurrency guarantees capacity for critical functions and avoids downstream throttling.
No extra charge applies to reserved concurrency.

For advanced performance optimization and cold-start reduction, consider combining reserved concurrency with provisioned concurrency.

References

Watch Video

Watch video content