Cloud infrastructure / AWS Fargate alternative

AWS Fargate alternative for persistent agent workers

AWS Fargate runs ECS tasks and EKS Pods without a customer-managed worker fleet. It supports durable mounted data through services such as EFS, while task-local storage is ephemeral. For a returning agent, distinguish files in those mounts from installed tools and configuration elsewhere in the container.

Start buildingCompare capabilities ↓

Reviewed 2026-10-19 · Published by the Sandbox0 team

When AWS Fargate fits

Choose Fargate for managed container scheduling integrated with ECS/EKS, AWS IAM and networking. Keep its application orchestration, deployment and service-integration controls when those are central to the system.

When to consider Sandbox0

Consider Sandbox0 for a returning compatible Linux worker whose complete writable RootFS retains tools and grows online. You still need to implement or retain container/service orchestration, autoscaling, load balancing and AWS integrations as required.

Capabilities

AWS Fargate and Sandbox0: what persists?

Feature comparison with documentation and verification status
Environment and runtime lifetimeAWS FargateRequested resources bill from image download until task termination. Retained EFS data has a separate lifecycle from the running task and its temporary root files.View evidence ↗Sandbox0ttl and hard_ttl default to 0 (disabled). Configure idle pause and deletion policies for the same persistent sandbox identity; an idle running sandbox still accrues compute charges.Read documentation →
Persistent filesAWS FargateEFS can retain mounted application data independently of a task.View evidence ↗Sandbox0The writable RootFS survives successful checkpoints, including software, system configuration, dependencies, and user data.Read documentation →
Whole RootFSAWS FargateTask-local storage is ephemeral; EFS does not establish whole mutable RootFS retention.View evidence ↗Sandbox0Persistence extends beyond /workspace to paths such as /usr/local, /etc, and /home. Runtime tmpfs and virtual filesystems are excluded.Read documentation →
Release compute, keep stateAWS FargateThe reviewed task storage contract does not establish arbitrary machine checkpoint/resume.View evidence ↗Sandbox0Pause compute between tasks; resume on available capacity. Default resume starts fresh processes from the saved filesystem.Read documentation →
Memory and process stateAWS FargateNo arbitrary Fargate process-memory checkpoint contract is established here.View evidence ↗Sandbox0Experimental memory checkpoints can preserve compatible process state. Use memory: true on both pause and resume on a compatible deployment. A filesystem checkpoint alone does not preserve running processes.Read documentation →

Online RootFS growth

On growth-enabled images, Sandbox0 automatically expands the writable RootFS as free space runs low, within a configured sparse logical ceiling up to 64 TiB. Agent processes keep running. Growth is enabled in the qualified Cloud region; confirm activation for other regions. Existing images and snapshots retain their captured storage geometry. The ceiling is logical capacity, not preallocated host disk.

Fargate supports larger task-local allocations and durable EFS mounts. These are different from an automatically expanding complete durable RootFS that retains system-path changes when an agent returns to the same environment.

Files while compute is paused

Sandbox0's Files API reads persistent files without resuming compute. Filesystem-only pauses also permit edits; retained-memory pauses are read-only and reject mutations with 409. SSH, SFTP, file watching, and runtime-only mounts require running compute.

EFS data can be accessed through that filesystem’s own clients while a task is stopped. That provides independent access to selected mounted data, not a full paused-task RootFS file interface.

See paused file operations →

Task-local storage and durable mounts differ

Current Fargate task storage starts at 20 GiB and can be configured up to 200 GiB. Container image layers use part of that allocation. EFS provides a separately managed filesystem mounted into containers; that durable path does not make all task-local system changes persistent.

Source 1 ↗Source 2 ↗

Resource requests determine the compute bill

Fargate bills requested CPU, memory and extra ephemeral storage, rather than only CPU instructions or the active RAM working set. Billing starts with image download and ends when the task terminates. Long image pulls and initialization are therefore part of execution cost.

Source 1 ↗

Minimums and architectures change effective cost

Linux is billed per second with a one-minute minimum, while Windows has a five-minute minimum. The published Linux/x86 US East example has different CPU/memory rates from Arm and other regions. Storage, logs, networking and public IPv4 charges have separate scope.

Source 1 ↗

Pricing

Compare billing models, then price your workload

CPU allocation, memory, billing minimums, disk, plan fees, and supported workload differ. These meters are not equal-performance machine quotes. Confirm region, resource units, and included allowances before estimating a total.

AWS Fargate

Pricing reviewed 2026-10-19

Linux/x86 CPU, US East N. Virginia
$0.000011244 per vCPU-second
Published regional reference; requested allocation and one-minute Linux minimum apply.
Rate source ↗
Linux/x86 memory, same region
$0.000001235 per GB-second
Requested memory, separate from CPU.
Rate source ↗
Additional task-local storage
$0.0000000308 per GB-second
Reference example above the included 20 GB pricing allowance; docs express task sizing in GiB.
Rate source ↗

Do not treat temporary storage as durable EFS or compare its price as retained RootFS storage. CPU architecture, operating system, region, Spot/commitment terms, external storage, logs and network costs affect the total.

Check current AWS Fargate pricing ↗

Sandbox0

$0.015/GiB-hour compute (billed by the second at approximately $0.000004167/GiB-second), plus $0.02/GiB-month of retained incremental RootFS and memory checkpoint data. Shared base images and platform recovery retention are excluded; paused sandboxes accrue storage charges only.

Compute / GiB-hour
$0.015
Retained data / GiB-month
$0.02
Paused compute
$0.00

A 4 GiB agent, 100 running hours

$6.00 compute

4 GiB × 100 hours, billed by the second. Keeping 20 GiB of billable incremental data for a full billing month adds $0.40. CPU is derived from memory by the platform; choose a tier that fits your workload.

Understand storage billing →

Before migrating from AWS Fargate

  1. Inventory task requests, image sizes, task-local writes, EFS mounts, IAM, network rules and service orchestration.
  2. Export required files and system setup into a compatible Sandbox0 template; preserve needed schedulers and AWS services.
  3. Test system-tool retention, online RootFS growth, paused files and total allocation, storage, initialization and network costs.

Common questions

Is Sandbox0 a drop-in AWS Fargate replacement?

Evaluate API compatibility and product scope. These products serve different core workflows, so a change may require a different architecture.

Does keeping files also keep a running agent?

Filesystem state and volatile process state are separate. Sandbox0 requires explicit compatible memory checkpoints to preserve running processes; its default filesystem resume starts fresh processes.

Sources and review scope

The Boat team's Battleships directory is the discovery source. References below describe the product and plan discussed on this page. Check current availability, regional terms and billing units before migrating. Where the cited material does not establish a capability or price, the comparison identifies that uncertainty.

Battleships research record ↗docs.aws.amazon.com/AmazonECS/latest/developerguide/fargate-task-storage.html ↗docs.aws.amazon.com/AmazonECS/latest/developerguide/efs-volumes.html ↗aws.amazon.com/fargate/pricing/ ↗Sandbox0 RootFS capacity →

Related comparisons

Akamai Cloud / Linode vs Sandbox0 →Alibaba Cloud ACK / Harbor backend vs Sandbox0 →Alibaba Cloud ECS vs Sandbox0 →Amazon Bedrock AgentCore vs Sandbox0 →AWS EC2 vs Sandbox0 →AWS Lambda Event Functions vs Sandbox0 →