How would SAFE network behave in such an event?.. both internal to the smaller network and then on the rejoining with the rest of the world?
Russia is considering whether to disconnect from the global internet briefly, as part of a test of its cyber-defences
If I recall correctly the earlier discussions of this, the two networks would function independently with no loss of data, but when they rejoined, conflicts (eg in Safecoin ownership) would be resolved in favour of the larger network.
I don’t know how that works, but I think that is what was anticipated.
Yes, that’s what I was expecting too but needs thinking through so the answer to such curiousities are robust… and also the case that user has access in some way to both networks.
I can’t quite see if it’s possible to have some address on the network that represents the coin and if that becomes inaccessible then it’s not spendable… because no one part of the real network can be relied on; so, unless there is an abstracted consensus on abstract location space that cannot be disrupted, it’s hard to see how problems with money are avoided.
Intelligent network needs to know the environment in which it sits?
I’d say like most big outages people would have to wait for the cutoff parts to rejoin so they can have complete access.
Remember though if one nation cutoff their internet then the chances that things like account info and random data will not be accessible is relatively small. For instance if 1/8 of the network is cutoff because the USA hits the stop button then a whole lot less than 1/8th of the small data sets will be inaccessible. (eg account info)
For large files where you need hundreds of chunks ALL to be there then the chance of files not being complete is not rare.
Yes, no worry for both small and large fragments carrying on and perhaps new data then merging simply when they rejoined but the double payment problem is different kind perhaps. For example, if SAFE was to host the BTC blockchain, it would need method to not get confusion from some split that is now in a class of deliberate State action, rather than rare theoretical oddity.
If the larger network takes precedence when they join back again then there is an issue for those in the smaller segment.
Also on a country cutting itself off then a lot of sections (in the smaller network) will break consensus and collapse and the disconnect since there is no more section. Then they will have to reconnect & make a “new” network and this is where work has to be done to ensure their data is not lost. And I gather this is one of the areas Maidsafe will have to consider in that process.
The larger segment will not lose too many segments and the problem is not so noticed and it should be able to recover and again the few sections lost in the large network will have to have their data recovered.
Maybe it will be a case of an effective merging where the remaining nodes keep going up the tree structure till there is enough nodes to form a section.
That’s a good answer!.. if the network is well distributed and then can sense damage across many sections, that could useful cause it to not recover; perhaps better to fail than perpetuate a small network?? Equivalent would be a global failure for the whole network being usefully applied on a smaller scale. Still I wonder the physical internet is too liable to a few cables but perhaps not so few that a whole continent - being a large fraction of the global, could ever fail. Single countries as large as Russia/China perhaps can be accommodated in this way.
If we take Australia as an example, we are really isolated geographically. New Guinea and New Zealand are our closest neighbours.
Yet the ability to cut us totally off due to cutting cables would require a significant amount of effort by a group with plenty of resources. We regularly have a cable cut and generally things just slow down. Many other countries have better interconnectivity than we do and a few have less. We still are a 3rd world country as far as connectivity goes.
Now thinking of a way to recover from large number of sections losing consensus.
- the remaining nodes realise that the section is broken
- they take their section number and remove the last bit and look for that section to join.
- continue this process of removing the last bit till a section is found
- then they join that section. This is where the code probably has to be written to ensure malice cannot happen
Now if no such section is found then the node assumes it is cutoff from any network and keeps looking for its section (with looking up the tree) till it finds it (connectivity re-established). This should work even for a home node or a set of nodes in a data centre, etc.
Cut/destroyed cables are not the biggest problem, today there is more than enough connections to make global internet last even in natural disasters. Problem is centralization of traffic in big datacenters and peering houses, where governments can put their killswitches. And also money driven decisions, when something could work, but somebody decides there is no money in it for him, so it doesnt.
This leads me to a question, how will the network react when the split is is only parcial?
- When something hapens and part of the network still has some connection to the rest, but the banwidth is very low.
- Something happens and there are some nodes is (lets say one country), they can comunicate with each other, but only few of them can comunicate also with tehe rest of the world (because they have private link to different country or something like that).
So may Russians make double spend during national disconnect ? I think that if all shops and exchanges will not hold their business, than they are in a big risk.