Good evening friends. This is a question regarding using Kafka to implement fault-tolerant messaging between distributed object-capabilities, Al la E. At issue is the around the bees’ knees approach to consuming an event once and deleting it out of all replicated partitions, immediately. This dodge is to somehow tell all relevant partitions that it is marked for deletion, then you have to request a partition compression to actually see the event removed from Kafka’s durable disk Log.
This is a real ObjectMemory. Could a mark and sweep eliminate such a marked-for-deletion event, more quickly?
My design is that ParrotTalk be as the secure session under & between Kafka services and vice-versa too. Remote Promises will talk to ParrotTalk which is a Kafka Session underneath, itself using a SecureSession. As RemotePromises publish / send DeliverMessages to an object in a different Vat, with a RemoteResolver registered back into the originating Vat stored into the RemoteObjectMemory (called the Scope of a RemotePromiseSession ), publishing into replicated partitions, consumed by the receiver, fault-tolerant messaging.
Compaction! That’s the correct term, is it not? Be well.
Have a Good One; Keep it, Light.
. .. … ‘…^,^
Sent from Callisto House :: decentralized mobile homeless solutions