Resource Discovery Standardization
SUMMARY - The following are the roles that need to be attributed to REDDnet members in order to provision the REDDnet with a resource discovery service that is interoperable with all the IBP-capable clients currently in use.
- Technical Information
- Communication interface (defining messaging format)
- Protocol specification (defining messaging content)
- Planning
- Timeline
- Development Resources
- Source of manpower
- Source of funding
BACKGROUND - A complete logistical networking system has four components:
- IBP-capable block virtualization agent
- storage node directory
- metadata repository
- system API
Given the distributed nature and flexibility of the technology, several models for taking advantage of logistical networking infrastructure are possible. There are two such models within the REDDnet community. The following are their two main differences:
1) metadata storage
- form - file v. database
- physical location - distributed/manual v. centralized/automated
2) storage consumption restrictions
- authority to make allocations - distributed v. centralized
Other than the possibility of impacting resource availability due to different initial assumptions, the first difference does not keep the two models from using the same infrastructure. Neither does the second as long as permission to perform allocations is restricted only by the client API rather than storage node configuration.
Unfortunately, the two models have been developed using separate resource discovery services. This has resulted in client-level incompatibility with each client unable to locate available storage resources without its respective discovery service. It is the desire of the REDDnet members to reestablish interoperability between these and future models by defining a simple, portable protocol to be used for resource discovery.