AWS Lambda Managed Instances with Java 25 and AWS SAM – Part 4 Monitoring, unsupported features, challenges and pricing

Source: Dev.to

![Cover image for AWS Lambda Managed Instances with Java 25 and AWS SAM – Part 4 Monitoring, unsupported features, challenges and pricing](https://media2.dev.to/dynamic/image/width=1000,height=420,fit=cover,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fk8wmjihn119mnm01mqpl.jpg)

[![AWS Heroes profile image](https://media2.dev.to/dynamic/image/width=50,height=50,fit=cover,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Forganization%2Fprofile_image%2F2491%2Ff0c1a659-c959-42cd-bb12-cd25909dd9db.png)](https://dev.to/aws-heroes)
![Vadym Kazulkin](https://media2.dev.to/dynamic/image/width=50,height=50,fit=cover,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F501061%2Ffa23b5ee-9c5f-48bd-b3bb-341a26a773c6.JPG)

## Introduction

In [part 1](https://dev.to/aws-heroes/aws-lambda-managed-instances-with-java-25-and-aws-sam-part-1-introduction-and-sample-application-1eb7) of the series we explained the ideas behind **AWS Lambda Managed Instances** and introduced our sample application.  
In [part 2](https://dev.to/aws-heroes/aws-lambda-managed-instances-with-java-25-and-aws-sam-part-2-create-capacity-provider-4cgo) we described what a **Lambda Capacity Provider** is and how to create it with AWS SAM.  
[Part 3](https://dev.to/aws-heroes/aws-lambda-managed-instances-with-java-25-and-aws-sam-part-3-create-lambda-function-with-lmi-1d9g) covered how to create Lambda functions and attach them to a capacity provider.

In this article we’ll cover:

* Monitoring  
* Currently unsupported features  
* Current challenges  
* Pricing  

---

## Capacity Provider Monitoring

The Capacity Provider exposes a set of CloudWatch metrics. You can view them by selecting the provider in the console and opening the **Monitoring** tab, or by browsing the metrics directly in CloudWatch.

![Capacity Provider monitoring screenshot](https://media2.dev.to/dynamic/image/width=800,height=&fit=scale-down,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2F6a5fy78zierltd6ofurt.png)

### Available metrics

- **Capacity provider CPU utilization**  
- **Capacity provider memory utilization**  
- **Capacity provider allocated utilization**  
- **Execution environment CPU utilization per function**  
- **Execution environment memory utilization per function**  
- **Execution environment count per function**  
- **Capacity provider instance counts**  

![Additional metrics screenshot](https://media2.dev.to/dynamic/image/width=800,height=&fit=scale-down,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fh6s1vw8rlga40pnsih93.png)

---

## Unsupported Features

The following Lambda features that are available with the default (Firecracker) compute type are **not supported** with Managed Instances:

* SnapStart  
* Provisioned concurrency  
* Reserved concurrency  

These features are designed to mitigate cold‑start latency in microVM‑based environments and therefore do not apply to pre‑provisioned EC2‑based LMIs.

### Instance‑type limitations

* No GPU support  
* No **t**‑family instances (t2, t3, t4g)  

For a full list of supported instance types, see the **AWS Lambda Pricing** page for your region.

---

## Current Challenges

* **Switching compute types** – You cannot seamlessly switch a function between the default Lambda compute model and Managed Instances. An easy toggle would be valuable because the default model is usually cheaper at low scale, while LMIs become cost‑effective at higher, steady‑state traffic.

* **Staging / test environments** – Setting up LMIs for low‑traffic or test workloads is cumbersome. The current workaround is to:

  1. Deploy the function with `min=0` and `max=0`.  
  2. When you need to run it, update the values (e.g., `min=1`, `max=2`).  
  3. After testing, revert to `min=0`, `max=0`.

  This requires multiple IaC redeployments, which slows down the workflow. It would be better if the minimum execution‑environment count could be set to 0 independently of the maximum, allowing the provider to scale down to zero automatically after a configurable idle period.

---

## Lambda Managed Instance Pricing

[Lambda Managed Instances pricing](https://aws.amazon.com/de/lambda/pricing/) consists of three components:

| Component                     | Description |
|-------------------------------|-------------|
| **Request charges**           | $0.20 per 1 M requests |
| **Compute management fee**    | 15 % premium on the on‑demand EC2 price for each instance type used by the capacity provider |
| **EC2 instance charges**      | Standard on‑demand EC2 pricing applies to the instances provisioned in your capacity provider. Savings can be achieved by using Compute Savings Plans, Reserved Instances, or Spot Instances where appropriate. |

> **Note:** The management fee is applied on top of the underlying EC2 cost, so the total cost per execution environment equals **EC2 price + 15 %**.

---  

*End of Part 4.*

## Savings Plans, Reserved Instances, or Other EC2 Pricing Options  

*Note:* Lambda Managed Instances functions **do not** incur separate charges for the execution duration of each request, unlike Lambda (default) compute‑type functions.

### Event Source Mappings  

For workloads that use provisioned **Event Poller Units (EPUs)** with event sources such as **Kafka** or **SQS**, the standard EPU pricing applies.

---  

## Conclusion  

In this article, we covered the following topics:

- Monitoring  
- Currently unsupported features  
- Current challenges  
- Pricing