Investigating storage protocols

OceanStore^?

http://oceanstore.cs.berkeley.edu/ -- This is a research project, not a protocol standard. Real use of ocean store is probably years out. It's pretty cool if it ever happens -- users hard drives are joined in an ocean of storage, data broken up into pieces, encrypted, and redundantly distributed. Not a real option today, and not a choice all users would be willing to make. It also doesn't necessarily solve any sharing issues.

Extensible Provisioning Protocol

http://www.ietf.org/rfc/rfc3730.txt - This is a new XML based IETF standard. It's really meant for provisioning registries, such as DNS registries. However, a provisioning protocol must naturally solve some of the same problems that a storage protocol must solve. The real defect from a Chandler point of view is a small (as yet) feature set.

xmlStorageSystem

This is Dave Winer's proposal for how to store blogs and "other stuff": http://www.soapware.org/xmlStorageSystem, based on SOAP 1.1. It's designed around uploading documents to a HTTP server (I believe it relies on HTTP for download entirely). Interestingly, it includes callbacks -- the user can subscribe to be notified via a new connection (to a specified port/hostname) that a file has been changed.

The only advantage of xmlStorageSystem is that it has notifications designed around HTTP document changes. Its disadvantages:

• It's massively under-specified, as is typical for a protocol designed mostly for one system. Error messages are not described at all.
• The notifications don't allow for proxies as XMPP notifications do.
• Change control is not in the hands of a standards body. Changes seem to be made without regard to version negotiation.

-- LisaDusseault - 02 Apr 2004