At the moment, off-chain storage is problematic as every node subscribed to a stream is automatically retrieving and writing to local storage all off-chain flagged data. This will clog up the network if only a 6-8 generalized streams are using in the blockchain network. To combat the inability to have granular publisher-based stream subscriptions and selective publisher-based off-chain data storage, one would need to create one stream per enterprise (and possibly have 100s of streams). Is this really the vision of streams? Please elucidate with your expertise, as this is not the architecture we envision...
