s3: Optimize WRITE_CHUNKSIZE to minimize an overhead
When processing the image data from a client, the previous and small
chunk size had a significant overhead. Let's be aligned with other chunk
size related parameters such as DEFAULT_LARGE_OBJECT_MIN_CHUNK_SIZE
(=5MiB) and DEFAULT_LARGE_OBJECT_CHUNK_SIZE (=10MiB).
The performance difference should be tangible especially with a
singlepart upload. And the upload time can be shortened for example as
follows:
Reviewed: https:/ /review. opendev. org/c/openstack /glance_ store/+ /799870 /opendev. org/openstack/ glance_ store/commit/ 32f9a1509bc94ba a6acaf508c922f3 b7edf5b65f
Committed: https:/
Submitter: "Zuul (22348)"
Branch: master
commit 32f9a1509bc94ba a6acaf508c922f3 b7edf5b65f
Author: Nobuto Murata <email address hidden>
Date: Thu Jul 8 01:36:38 2021 +0900
s3: Optimize WRITE_CHUNKSIZE to minimize an overhead
When processing the image data from a client, the previous and small LARGE_OBJECT_ MIN_CHUNK_ SIZE LARGE_OBJECT_ CHUNK_SIZE (=10MiB).
chunk size had a significant overhead. Let's be aligned with other chunk
size related parameters such as DEFAULT_
(=5MiB) and DEFAULT_
The performance difference should be tangible especially with a
singlepart upload. And the upload time can be shortened for example as
follows:
100 MiB: 36.5s -> 4.0s
200 MiB: 2m16.8s -> 6.1s
300 MiB: 4m50.4s -> 9.1s
Closes-Bug: 1934849 f7ffa76bb7ca63d 8d112036b70
Change-Id: Icecac80dd9e4e9