Really really basic stuff

[tfa]

@digipl

for now the calculation of resource available are based in the concept of “sacrificial data”

Unfortunately sacrificial data have been removed a long time ago (the low balance error only indicates that a client has run out of put credits).

Without this information I don’t know how a section can guess that a node is full (I mean reliably).

[Traktion]

Great post! I suspect the algorithm could be improved over time for these sort of things.

[mav]

Great ideas @happybeing. I agree that using reported size could probably be exploited.

I think my assumption that farming rate is meant to balance the provision of supply is incorrect.

The network probably shouldn’t care if there’s massive oversupply of capacity, which is a waste but not a problem for the network; it’s only a problem for vault operators (wasted cost electricity etc to maintain unused storage space).

On the flip side, the network should really care if there’s an imminent shortage of supply. That could cause a lot of damage in lots of different ways.

So the rate algorithm isn’t about balancing, it’s about provisioning.

I think this means the important metrics feeding into farm rate will be full vaults, ‘overflow’*, churn rates.

But ‘available resources’ won’t be an important metric.

* not sure if there’s a word for this yet, but when a chunk goes to a part of the network with lots of full vaults, how much shuffling of chunks has to happen to store that new chunk.

Also i’m in favour of these posts being moved to a separate topic, it’s not ‘really really basic stuff’, it just happened to be in the same topic as the post I was responding to.

[happybeing]

The reason the network may care about oversupply is efficiency and cost of storage. So if there are too many vaults with too much free storage, that’s a signal it can pay farmers less and offer cheaper storage.

At least that’s how I see it but it’s hard to model these things!

[mav]

@happybeing

it can pay farmers less and offer is cheaper storage

ah yes of course the rate affects both farmers and consumers. Such a fascinating and complex topic. I’m so excited for when maidsafe starts publishing more work on the farming algorithm!