Delayed Resume

Builds on Common Methodology

Times reading data from a sandbox that was paused or idle for 0 s, 60 s or 15 min.

Interactive coding agents, chat workspaces, and notebook sessions sit idle between user turns. Pausing a sandbox between turns stops compute billing while preserving the user's workspace, but only if the sandbox can pause and wake back up with its files intact.

Furthermore, a provider can return from a resume API call before the sandbox's data has been paged back in, shifting the real wait onto the first command that touches the data.

The Pause and Resume tab measures both sides of that lifecycle: Pause and Resume tests immediate pause, resume, and data read-back across memory and disk loads, while Delayed Resume tests wake and read time after a sandbox has been paused or left idle across selected wait durations.

View the Procedure View Delayed Resume Results Download Raw Data View the Parameters

Metrics

Delayed Resume is the sum after the wait of two timed stages: Resume (resume for explicit-pause providers, or 0 for auto-idle providers that wake on command) and Read (reading back the 4096-byte marker file with dd).

Timed Part	Starts	Ends
Resume	The resume request	The resume call returns
Read after wait	The resume, or the first read where the provider has no resume call	The marker and clock readings return to the job host

Measurement

Each trial uses a sandbox created from BoxRouter's pinned standard image.

1. Create sandbox, write 4096-byte marker, read it once (control), call pause if supported, and wait 0 s, 60 s, or 15 min. For auto-idle providers, the sandbox is left untouched for that wait so it can enter standby on its own.
2. Resume to Data Read Call resume where supported (Resume). Timed
3. Read the 4096-byte marker file again with dd (Read). Timed
4. Verify standby duration and destroy sandbox. The harness compares guest wall-clock time against /proc/uptime to verify whether the kernel was frozen during standby.

The 4096-byte marker file is generated from SHAKE256 (seed = 20261004), and 15 min is the default headline view. The three wait durations (0 s, 60 s, and 15 min) test whether a paused or idle sandbox is evicted from local host memory to colder network storage as idle time grows.

Scope Limits and Caveats

• Waits longer than Common Methodology's idle timeout (the 15 min variant) may include a new TCP and TLS handshake in the post-wait call, and auto-idle providers that fail to enter standby during a positive wait have their metric withheld.
• Blaxel sandboxes have no explicit pause; they auto-idle after a brief inactivity window. read includes the wake from idle.
• Vercel implements stop and resume by snapshotting the filesystem on stop and booting a new VM from that snapshot on resume.

Failures and Exclusions

Outcome classifications, top-up rounds, and outlier rules follow Common Methodology; the first failed step ends the trial without in-trial retries.

A trial succeeds only when all steps complete within timeout, the post-wait marker read matches the pre-wait SHA-256 digest, and the sandbox is confirmed destroyed. On positive wait variants (60 s and 15 min), an auto-idle provider whose guest /proc/uptime kept advancing throughout the wait (never entering standby) has its metric withheld.

Step	Time Limit
Create	5 min
Read before wait	5 min
Pause	5 min
Idle wait	15 min
Resume	5 min
Read after wait	5 min
Destroy	5 min

Statistics

Based On

Common Methodology

Per-unit percentiles, bootstrap confidence intervals, and tail rules follow Common Methodology.

A provider is ranked on a wait variant once it has at least 18 successful trials.

Sandbox and Trials

Image Digest

sha256

Image Version

0.3.11

Image Registry Page

View on the Registry

Pull the Image

docker pull ghcr.io/boxrouter/standard-image@sha256:89d4c97048dbdfb8b6d85ad91833962ff38d1c334a5f47e6de80a9762310191b Copy

Lifetime

25.0 min

Idle Connection Time

5.0 min

From Common Methodology

Idle Connection Exceptions

Provider	Rule	File Transfer Idle Time
Daytona	The SDK has no idle-timeout setting. Its command channel keeps its connection open, and its file-transfer channel reconnects after 5 s of idle time.	5.0 s

From Common Methodology

Network Path Attempts

10

From Common Methodology

Mode

Single

Sandboxes per Trial

1

Parallel Trials

10

Concurrent Sandboxes

1

At a Glance

Quantity

Time

Per

Round

Unit

Milliseconds (ms)

Direction

Lower Is Better

Statistic

Median

Clock

Job-host clock

Machine Size

2 vCPU / 4 GiB

Trials

20

Variants

• Resumed at once
• Resumed after 60 s
• Resumed after 15 min