Replacing current NetarchiveSuite Archive module
Under development
A number of new technologies have appeared which makes it possible to replaced the current archive codebase with standard components.The current archive functionality can be split into the following 3 areas:
Wayback Access
Current solution
The current solution is to fetch resources through jms messages to the bitarchives.
Pros:
- Read access is handled the same way as write access, and the same model for logging and autherization can be used.
- Both replicas can be used for retrieving data (can they).
Disadvantage:
Alternate solution
Switch to let Wayback access files directly though mounted RO drives.
Pros:
- Wouldn't put stress on the message broker or bitarchive infrastructure.
- Wouldn't depend on the message broker or bitarchive infrastructure scalability.
Disadvantage:
Massprocessing
Current solution
This is currently implemented through the BatchJob functionality, which has been develop
Pros:
- Full control of source code
Disadvantage:
- All improvements have to be done with NetarchiveSuite ressources.
- Somewhat unstable.
Alternate solution
The de facto standard platform for massprocessing is Hadoop, which is used at an increasingly number of webarchiving institutions for analysing the stored web data. Both SB and KB have established Hadoop clusters, which are already used for processing the Netarkivet.dk archive.
Pros:
- Hadoop is an mature standard processing platform for large datasets.
- Comes with a huge set of tools, including Webarchiving tools.
- Very robust and scalable.
- Enables processing resources and data to be separated (SB setup)
Disadvantage:
- Migration cost
Bitpreservation
Current solution
Pros:
Cons:
Alternate solution
Switch to using a Bitrepository system for the Bitpreservation
Pros:
Disadvantage: