Coordinating Node RequirementsΒΆ
This list of requirements for Coordinating Node functionality was developed during a brainstorming session at the 20090602 ABQ VDC-TWG meeting.
Todo
Estimate some times for the unknowns here.
- IDs not discoverable until all CNs have confirmed copy
- Metadata searches return in max xxx secs
- CN can store and search > xxx metadata records
- CN can store and search multiple metadata standards - specifically (list Metadata_formats_of_interest.txt)
- CN can load-balance to maximize performance
- Any xxx CNs can be offline without affecting D1 services
- xxx concurrent transactions can be supported - different for different types of transaction e.g. delete vs. insert
- Each CN should be available xxx% time
- CN should expose MD in search within xxx secs of insert by MN
- CN services should be designed to be independently scalable
- CN services should be geographically replicated
- CN have complete metadata copies from all MNs
- D1 should be TRAC and/or DRAMBORA trusted repositories
- CN should respect replication policies of MNs
- CN should ensure that any given data set is available at any time (subject to policy restrictions) even when xxx% of MNs are offline
- CNs must validate that: - data are available at all replicas - replicas are valid
- CNs ensure that data are available in current formats (data format upgrade?)
- CNs should be economical to run and maintain
- CNs should be able to be remotely administered
- CNs should be secure and deflect malicious intent
- CNs should provide services to detect rogue data (viruses, copyright, )
