I read that there will be updates to the SAFE network after the initial version is launched. I was interested to know how these updates would occur, i.e. how would the network update be introduced and propagated to the individual nodes without there being some sort of central update node/server?
Many thanks!
The specifics are still to be determined in test network 4, but if my memory is correct the plan is for every vault to decide to upgrade or not on its own, it would run both the old and new version and see which is more efficient.
Though personally I don’t know if it’s gonna be possible without turning each vault into a GAI.
Presumably through the network itself as immutable data objects. Nodes could poll an AD, sort of like an RSS feed, to see when new releases are available.
As drehb said there would likely be polling a data object.
Data becomes the central authority.
Your notes/diary/whatever you keep are uploaded to SAFE as the central source for your notes. So too would be for updates. No longer is it the server/node/human authority, but data. This is the essence of a decentralised network holding your data.
so if I understand correctly:
the update will be pushed to all nodes, and then there will be a system that will look at the efficiency of those two version and decide which is better.
is it gonna be determined by an algorithm? like test one version with some PUTS and GETS and then the second one and the one that rewards the most to vaults and costs less per PUT will win?
I think if that is true it might not be good (I know my take out is simple) cause that might not take into consideration about safety and redundancy and other factors that are very important
so to sum up what will be the way of choosing the best version of the safe network by the vaults? (you say data I dont understand that part)
Speculation warning!!!
I expect vaults will ship with an algorithm which runs both current and updated vaults side by side, and makes some assessment on whether and when to adopt the upgrade.
This might include the ability to check first with the user, or to automatically accept upgrades as is common with operating system and so updates.
This seems sensible because it is really no different than making sure that any vault is doing what the network requires for proper functioning.
The question is though, how can this be used to attack the network? I think the job of the algorithm will be to prevent this.
I’d expect that nodes would pull the update
@happybeing is right, at this time it is mostly speculation as to the actual mechanics
What we do know is that the network will have a mechanism to update.