[
https://issues.apache.org/jira/browse/OAK-10418?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17759008#comment-17759008
]
Julian Reschke commented on OAK-10418:
--------------------------------------
I don't quite get how this helps unless the main improvement is to actually
separate the upload process from the migration step?
> Query : Faster binaries migration to azure cloud while using OAK repo
> ---------------------------------------------------------------------
>
> Key: OAK-10418
> URL: https://issues.apache.org/jira/browse/OAK-10418
> Project: Jackrabbit Oak
> Issue Type: Documentation
> Components: jackrabbit-api
> Reporter: Bharat Wadhwa
> Assignee: Thomas Mueller
> Priority: Minor
> Labels: jackrabbit
>
> The Content Transfer Tool is a tool developed by Adobe that you can use to
> initiate the migration of existing content from a source AEM instance
> (on-premise or AMS) to the target AEM Cloud Service instance.
> For migrating inline blobs, we use the OAK API, which helps upload this data
> to an Azure container. As per the current architecture, it spawns 20 threads,
> with each thread uploading one blob at a time. [Repo
> link|[http://example.com|https://github.com/apache/jackrabbit-oak/blob/trunk/oak-blob-cloud-azure/src/main/java/org/apache/jackrabbit/oak/blob/cloud/azure/blobstorage/AzureBlobStoreBackend.java#L290]]
> However, uploading one blob at a time introduces network latency with each
> call, ultimately increasing the overall migration time.
> Proposed solution: Rather than uploading the blobs one at a time, we would
> store them in temporary storage. Once that step is complete, we can utilize
> the publicly available azCopy feature, which significantly accelerates the
> data migration process.
> I conducted a Proof of Concept (POC) locally, and we observed a 15%
> improvement with this approach.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
