Open gnott opened 6 years ago
Looking back at some notes, we were also thinking about a separate folder for PoA and VoR articles.
Also, the folder / path structure is not decided. The zip file name format is also not decided.
In general, the fewer buckets an application has, the easier to maintain as the paths are kept within the application rather than being exported to infrastructure definitions in other project. For example, a single bucket with poa/
and vor/
prefixes (or other prefixes, no preferences expressed here) is functionally identical to two different buckets but preferable to multiple ones because it's a single infrastructure definition that will change less often than multiple buckets. S3 has filters by prefix so listing the files (the only bucket-wide operation) can still be performed.
On that note, we have prod-elife-published
which is publicly readable and also has a CDN on top. The CDN could also be useful to deliver these files so they could be pushed there as zip/poa/...
/zip/vor/...
(again, just sample prefixes, no preferences).
Super suggestions Giorgio, thanks. Reusing the existing bucket should be fine. It sounds like we want to keep poa
and vor
content separate, so the zip/poa/
and zip/vor/
folders are perfect.
The goal is to create public readable bucket(s) to store article zip files for anyone to download.
Some initial ideas are not finalised yet, for example
Should we have multiple buckets?
@giorgiosironi perhaps you may be able to help here if the buckets should be defined in infrastructure definitions.