Most people make the decision based on functional and privacy requirements, not resource consumption.

See functional and privacy considerations here:
https://hyperledger-fabric.readthedocs.io/en/latest/private-data/private-data.html#when-to-use-a-collection-within-a-channel-vs-a-separate-channel

For example
- If a piece of data is shared among a fixed sets of orgs forever, and never transferred to other orgs, then channels make sense for complete data isolation among the orgs.
- If you want to exchange assets among many organizations, keep asset data private among the current asset owners, but be able to verify history upon a transfer, then a single channel with multiple private data collections (perhaps one per org) would make sense. This is the scenario for the newest Fabric sample: https://github.com/hyperledger/fabric-samples/tree/master/asset-transfer-secured-agreement/chaincode-go


Dave Enyeart

"Santiago Figueroa Lorenzo" ---06/13/2020 12:30:42 PM---Hi fabric experts, I am trying to establish a comparison of the advantages of using private

From: "Santiago Figueroa Lorenzo" <santiagofigueroalorenzo@...>
To: fabric@...
Date: 06/13/2020 12:30 PM
Subject: [EXTERNAL] [Hyperledger Fabric] [Fabric] Comparison private data collection vs channel
Sent by: fabric@...

---

Hi fabric experts,

I am trying to establish a comparison of the advantages of using private data collections over increasing the number of channels. I have been thinking about measuring how much additional resource is consumed in a node as the number of channels increases, compared to the resource consumption by increasing the number of collections that are stored in the private state DB.

I have tried to find some prometheus metrics that would allow me to make another comparison, but I have not found any that can be used for that purpose. I'd appreciate it if anyone has any ideas about that.

Thank you in advance,

Santiago.