Both VRAM and DISK provide smart contract storage with different characteristics:

1. data capacity
2. read/write latency
3. data replication (redundancy)
4. complexity

This article correctly identifies that VRAM and DISK may offer different levels of data capacity vs latency. Have you tested what the trade-off is between these two storage mechanisms are — aka what is the capacity/cpu tradeoff for VRAM and DISK?

This article also correctly identifies that the data replication level is able to be controlled externally through the VRAM system, whereas all data on DISK is replicated on every node.

It is also not clear how VRAM and DISK behave differently with IPFS. AFAIKT both can be used to store an IPFS hash to the same effect (given the different characteristics of each) so why are you bringing it up?

There is a forth dimension to choosing storage mechanisms which is the complexity. As DISK will be built into the blockchain nodes, reliance on third-party software/infrastructure will be unnecessary which will provide additional advantages for developers. How does Liquid Apps view this?

When the trade-off between data capacity and latency is very similar between VRAM and DISK I think most developers will choose DISK because it will be easier to use, have less external dependencies and have better stability and documentation from B1. Without having done tested this and with my incomplete but also fair understanding of VRAM and DISK, I imagine the trade-off is quite similar so I would really like to know if you have calculated this and your thought.

Thanks!