Orderer ledger pruning
Our company (norbloc AB) is facing a problem with managing disk space occupied by ledger DB. Thanks to the snapshot feature in Fabric v2.3 the problem was solved for the peers, but we still have the problem with orderers - the disk space consumption is growing uncontrollably.
This leads us to the necessity of having ledger pruning capabilities for orderers.
This email is basically an indication of intent. We would like to find if there is (or was) already an effort to implement orderer ledger pruning as we would like to avoid duplication of effort. For example Brett Logan mentioned in RocketChat that there was some interest from Fujitsu on the topic, so we would like to know what is the status of that effort.
In case we find that there is no ongoing work on the topic, we intend to start the RFC process, at the same time developing an initial reference codebase.
Our initial approach would be basically to implement the peers' snapshot and "join from snapshot" functionality in orderers. We understand that this might sound a bit naive, but we have to start from somewhere. We would greatly appreciate it if more knowledgeable members of the community point to us the biggest challenges of this approach and in general what are the considerations we should keep in mind while designing this.
As a side note - we are also quite interested in the implementation of https://jira.hyperledger.org/browse/FAB-8007 as it promises 50-to-66% size reduction of the block ledger DB.
// Vitalii Demianets @ norbloc AB