- Easy to implement
- Easy to describe (naming etc.)
- Impossible to span very large files over multiple containers
- Some containers can only get so big (e.g. ZIP 4GB)
- Possible to span very large files over multiple containers
- Allows users to choose the best format for their specific use case
- More possibilities to store OCFL objects in storage systems
- Easier to transfer objects over network
- More complex to implement
- More complex to describe (naming etc.)
- More complex to handle (e.g. copy, move, validate, ...)
- Might require manual partitioning of containers (see next section)
Supporting multipart containers per version can be done in different ways:
- Independent containers
- Every container is a fully usable container which can be used or packaged independently
- Content can be split logically (e.g. by folders) or physically (e.g. by size) into containers (maybe define that as an extension)
- Binary container split
- Container is split into sub-containers at binary level and must be concatenated to get a usable container
- Can only be split by size
Type | Pros | Cons |
---|---|---|
Independent Containers | Works with all formats | Potential overhead |
Easy to implement | Might need to split containers manually / with external programs | |
Usable in any order | Hard to define clearly, might end up in individual implementations | |
Can retrieve a single sub-container | ||
Clear overview over file location | ||
Binary Container Split | Works with all formats | Some formats require splits to be combined for extraction |
Native support if allowed by format | One corrupt sub-container can break the whole container | |
Complexity depends on format | ||
Not all formats support this |