The ExecutionTiming object

The ExecutionTiming object provides detailed telemetry about how a job was processed. It is available on the Job object as execution once the job has started processing.

Use execution timing to monitor encoding performance, optimize your pipeline, and debug slow jobs. All duration fields are in milliseconds.

Attributes

Attribute	Type	Description
`instance_id`	string \| null	Unique identifier of the worker instance that processed the job.
`instance_type`	string \| null	Machine type of the worker (e.g., `c3-standard-8`, `n2-highcpu-16`).
`instance_location`	string \| null	Cloud region of the worker (e.g., `us-central1`, `europe-west1`).
`vcpu_count`	integer \| null	Number of vCPUs available to the worker.
`memory_mb`	integer \| null	Memory available to the worker in megabytes.
`boot_duration_ms`	integer \| null	Time to boot the worker instance.
`download_duration_ms`	integer \| null	Time to download the input file.
`probe_duration_ms`	integer \| null	Time to probe (analyze) the input file.
`encode_duration_ms`	integer \| null	Time spent encoding all outputs. This is typically the longest phase.
`upload_duration_ms`	integer \| null	Time to upload all output files.
`packaging_duration_ms`	integer \| null	Time to package streaming outputs (manifest generation, segment finalization).
`total_duration_ms`	integer \| null	Total wall-clock time from start to finish.
`download_bytes`	integer \| null	Input file size downloaded in bytes.
`upload_bytes`	integer \| null	Total output size uploaded in bytes.
`download_speed_mbps`	number \| null	Average download speed in megabits per second.
`upload_speed_mbps`	number \| null	Average upload speed in megabits per second.
`avg_cpu_percent`	number \| null	Average CPU utilization during encoding (0—100).
`peak_cpu_percent`	number \| null	Peak CPU utilization during encoding (0—100).
`avg_memory_mb`	integer \| null	Average memory usage during encoding in megabytes.
`peak_memory_mb`	integer \| null	Peak memory usage during encoding in megabytes.
`chunk_count`	integer \| null	Number of chunks the input was split into for parallel encoding.
`chunk_strategy`	string \| null	Chunking strategy used (e.g., `scene-detect`, `fixed-duration`, `none`).
`exit_code`	integer \| null	Process exit code. `0` indicates success.
`exit_reason`	string \| null	Human-readable exit reason if the process terminated abnormally.

{
  "instance_id": "wrk_8f3a2b1c4d5e",
  "instance_type": "c3-standard-8",
  "instance_location": "europe-west1",
  "vcpu_count": 8,
  "memory_mb": 32768,
  "boot_duration_ms": 850,
  "download_duration_ms": 3200,
  "probe_duration_ms": 1150,
  "encode_duration_ms": 42300,
  "upload_duration_ms": 5600,
  "packaging_duration_ms": 1200,
  "total_duration_ms": 54300,
  "download_bytes": 245366784,
  "upload_bytes": 178520064,
  "download_speed_mbps": 613.4,
  "upload_speed_mbps": 255.0,
  "avg_cpu_percent": 92.3,
  "peak_cpu_percent": 99.8,
  "avg_memory_mb": 4096,
  "peak_memory_mb": 6144,
  "chunk_count": 4,
  "chunk_strategy": "scene-detect",
  "exit_code": 0,
  "exit_reason": null
}

Phase durations

The encoding pipeline consists of sequential phases. Each phase’s duration is tracked independently:

boot --> download --> probe --> encode --> upload --> packaging
                                                        |
                                                total_duration_ms

Phase	Field	Description
Boot	`boot_duration_ms`	Worker instance startup time. For warm workers, this is near-zero.
Download	`download_duration_ms`	Time to download the input file from the source origin or URL.
Probe	`probe_duration_ms`	Time to analyze the input file and extract metadata (codec, resolution, duration, etc.).
Encode	`encode_duration_ms`	Time spent transcoding all output variants. This is typically 80—95% of total time.
Upload	`upload_duration_ms`	Time to upload all generated output files to the output origin.
Packaging	`packaging_duration_ms`	Time to generate streaming manifests and finalize segment files. `0` for progressive outputs.

total_duration_ms is the wall-clock time and may be slightly larger than the sum of individual phases due to overhead between phases.

Transfer statistics

Attribute	Description
`download_bytes`	Total bytes downloaded. Matches `input_metadata.size_bytes`.
`upload_bytes`	Total bytes uploaded across all outputs.
`download_speed_mbps`	`download_bytes * 8 / download_duration_ms / 1000`.
`upload_speed_mbps`	`upload_bytes * 8 / upload_duration_ms / 1000`.

Resource usage

Attribute	Description
`avg_cpu_percent`	Average CPU usage across all cores during the encode phase. Values above 90% indicate efficient utilization.
`peak_cpu_percent`	Maximum CPU usage recorded during encoding.
`avg_memory_mb`	Average memory footprint during encoding.
`peak_memory_mb`	Maximum memory usage. High values relative to `memory_mb` may indicate the job needed a larger instance.

Chunk encoding

For long-duration inputs, Transcodely may split the video into chunks for parallel encoding:

Attribute	Description
`chunk_count`	Number of parallel chunks. `1` means no chunking was used.
`chunk_strategy`	How chunks were determined. `scene-detect` splits at scene boundaries for seamless concatenation. `fixed-duration` splits at fixed intervals. `none` means the entire file was encoded as one unit.

Availability

Job Status	`execution`
`pending`	`null`
`probing`	Partially populated (worker details, boot/download/probe times).
`awaiting_confirmation`	Partially populated.
`processing`	Progressively populated as phases complete.
`completed`	Fully populated.
`failed`	Populated up to the phase where failure occurred.
`canceled`	Populated up to the phase where cancellation occurred.

#The ExecutionTiming object

#Attributes

#Phase durations

#Transfer statistics

#Resource usage

#Chunk encoding

#Availability