Parallel composite uploads

One strategy for uploading large files is called parallel composite uploads. In such an upload, a file is divided into up to 32 chunks, the chunks are uploaded in parallel to temporary objects, the final object is recreated using the temporary objects, and the temporary objects are deleted.

Parallel composite uploads can be significantly faster if network and disk speed are not limiting factors; however, the final object stored in your bucket is a composite object, which only has a crc32c hash and not an MD5 hash. As a result, you must use crcmod to perform integrity checks when downloading the object with Python applications. You should only perform parallel composite uploads if the following apply:

  • Any Python user, including gsutil users, who needs to download your objects has either google-crc32c or crcmod installed.

    For example, if you use Python to upload video assets that are only served by a Java application, parallel composite uploads are a good choice because there are efficient CRC32C implementations available in Java.

  • You do not need the uploaded objects to have an MD5 hash.

How tools and APIs use parallel composite uploads

Depending on how you interact with Cloud Storage, parallel composite uploads might be managed automatically on your behalf. This section describes parallel composite upload behavior for different tools and provides information for how you can modify the behavior.

Console

The Google Cloud console does not perform parallel composite uploads

Command line

You can configure how and when gcloud storage cp performs parallel composite uploads by modifying the following properties:

  • storage/parallel_composite_upload_enabled: Property for enabling parallel composite uploads. If False, disable parallel composite uploads. If True or None, perform parallel composite uploads for objects that meet the criteria defined in the other properties. The default setting is None.

  • storage/parallel_composite_upload_compatibility_check: Property for toggling safety checks. If True, gcloud storage only performs parallel composite uploads when all of the following conditions are met:

    Note that in order to check these conditions, the gcloud CLI retrieves the metadata for the destination bucket as part of the upload command.

    If False, gcloud storage does not perform any checks. The default setting is True.

  • storage/parallel_composite_upload_threshold: The minimum total file size for performing a parallel composite upload. The default setting is 150 MiB.

  • storage/parallel_composite_upload_component_size: The maximum size for each temporary object. The property is ignored if the total file size is so large that it would require more than 32 chunks at this size.

  • storage/parallel_composite_upload_component_prefix: The prefix used when naming temporary objects. This property can be set either as an absolute path or as a path relative to the final object. See the property description for more information. The default prefix is the absolute path /gcloud/tmp/parallel_composite_uploads/see_gcloud_storage_cp_help_for_details.

You can modify these properties by creating a named configuration and applying the configuration either on a per-command basis by using the --configuration project-wide flag or for all gcloud CLI commands by using the gcloud config set command.

No additional local disk space is required when using gcloud CLI to perform parallel composite uploads. If a parallel composite upload fails prior to composition, run the gcloud CLI command again to take advantage of resumable uploads for the temporary objects that failed. Any temporary objects that uploaded successfully before the failure do not get re-uploaded when you resume the upload.

Temporary objects are named in the following fashion:

TEMPORARY_PREFIX/RANDOM_VALUE_HEX_DIGEST_COMPONENT_ID

Where:

  • TEMPORARY_PREFIX is controlled by the storage/parallel_composite_upload_component_prefix property.
  • RANDOM_VALUE is a random numerical value.
  • HEX_DIGEST is a hash derived from the name of the source resource.
  • COMPONENT_ID is the sequential number of the component.

Generally, temporary objects are deleted at the end of a parallel composite upload, but to avoid leaving temporary objects around, you should check the exit status from the gcloud CLI command, and you should manually delete any temporary objects that were uploaded as part of any aborted upload.

REST APIs

Both the JSON API and XML API support uploading object chunks in parallel and recombining them into a single object using the compose operation.

Keep the following in mind when designing code for parallel composite uploads:

  • When using the compose operation, the source objects are unaffected by the composition process.

    This means that if they are meant to be temporary, you must explicitly delete them once you've successfully completed the composition, or else the source objects remain in your bucket and are billed accordingly.

  • In order to protect against changes to source objects between the upload and compose requests, you should provide an expected generation number for each source.