Objective
Learn how to optimize the sending of your files to your bucket.
Requirements
- awscli installed
Instructions
The aws s3 transfer commands are multithreaded. At any time, several requests are in progress.
For example, if you download a directory via aws s3 cp localdir s3://bucket/ --recursive, the AWS CLI might download localdir/file1, localdir/file2, and localdir/file3 in parallel. The max_concurrent_requests value specifies the maximum number of transfer commands allowed at any single time.
You may need to change this value for several reasons:
- Decrease this value
On some environments, the default value of 10 concurrent requests can overwhelm a system. This can cause connection interruptions or slow down system responsiveness. By lowering this value, S3 transfer commands will be less resource-intensive. The downside is that S3 transfers may take longer to complete. It may be necessary to reduce this value if you are using a tool such as trickle to limit bandwidth.
- Increasing this value
In some scenarios, you may want S3 transfers to complete as quickly as possible, using as much bandwidth as necessary. In this scenario, the default number of concurrent requests may not be sufficient to use all available network bandwidth. Increasing this value can improve the time required to complete an S3 transfer.
This value should be set under the top-level s3 key in the AWS configuration file, whose default location is ~/.aws/config.
user@host:~$ cat ~/.aws/config
[profile default]region= <region_in_lowercase>
s3=endpoint_url= https://s3.<region_in_lowercase>.io.cloud.ovh.us
max_concurrent_requests= 20
signature_version= s3v4
s3api=endpoint_url= https://s3.<region_in_lowercase>.io.cloud.ovh.us
You can also set this value using the aws configure set command.
aws configure set default.s3.max_concurrent_requests 20