Migration of Mannequin Assets from Archive
Overview
Merch are going to be reviewing and updating the image sets on PDPs in the coming months and would like to be able to update / delete / reorder existing assets in Cloudinary.
For assets that were migrated from Amplience to Cloudinary, this is not currently possible because there is no link between Hybris and Cloudinary for those assets.
To add more assets to an existing set that was added via Amplience, Merch could simply add the additional assets to Cloudinary and use the standard Hybris sync
- To replace the assets from an existing set that was added via Amplience, Merch would need to go to Hybris and delete the assets and then add the new assets via Cloudinary
- It is not possible for Merch to reorder existing assets that were added via Amplience, and it is not possible for them to reorder image sets that are a mix of assets added via amplience and via cloudinary
- Merch wanted access to assets that were migrated from Amplience in order to reupload them and use them to update image sets - we have given them read only access to the archive folder
We discussed writing a script to move assets from the archive folder into the mannequin folder, to prevent Merch from having to re-upload assets and create duplication across the archive and mannequin folders.
Initial assets we've identified in Archive
- 2585 SKUs which equates to 13672 assets.
Pros, Cons and Risks
Pros
- 13672 assets moved from archive
- This reduces the need for Merch or Brand to copy images from archive into the flat folder structure
- Reduces the changes of browser specific images being moved into the flat folder structure. AVIF is a prime example
- Reduces the number of assets in archive down to 12k
Cons / Risks
- We can only migrate assets that are named correctly
- This could turn into an ongoing process. We can continue to find tune the assets that we can rename in archive in different rounds.
- We cannot take responsibility for PDPs that may be missing images after migration. As we have no automated testing we won't know what has been impacted
- The biggest risk is stories. Stories contain the CMS Paragraph Component which adds free text into the DB. This means that we may have mannequin urls in there. This would require us to do a DB find and replace.
- This again can impact the life of the squad as it would turn into an ongoing process
- Cannot be tested accurately using rapha-test and UAT as it is not a true representation of Production.
- We have to performance test as we may run into concurrency issues. We know there are limitations on how Hybris can sync between catalogs.