Unless you mean this:
Instead of nodes holding the key only in RAM they write it to disk. On restart, they read and republish the data?
2 Likes
Leaving your PC on 24/7 is the greenest solution. Nothing kills electronics, especially PSUs like power-cycling. Weigh up the cost of a PC idling(with full power saving) while you sleep vs the cost of a new PSU every couple of years.
6 Likes
If there was no risk of an attack, this could be an excellent solution. 
Or perhaps in terms of critical data (e.g., you once blogged about how it would be impossible for autonomous cars to move without a network like SafeNet), a system could be set up to tag critical data with additional redundancy using AI-based file deletion/replication analysis? 
2 Likes
It is not that simple. Disconnecting PC from power totally is not good, but it is fine to put PC to sleep or shutdown.
In general biggest killer is heat which degrades electrolytic capacitors in time. This than makes devices more prone to failure on power on/off or power fluctuations.
4 Likes
Correct. I was being brief - for once
I think the greatest chance of something like this happening would be in the early stages of the network after some calamity (earthquake, hurricane, wildfire, etc.) affected a wide geographical area and encompassed all the nodes holding the same data. That seems far-fetched, but early-on could happen. The way to mitigate this possibility, of course, would be to increase the likelihood that nodes holding the same data are geographically dispersed. Is there a way to do this, especially during the first year or so?
Such an event, even though unlikely, could do irreparable harm to the project, mostly from the attendant negative publicity.
1 Like
Not for all applications, some current cryptocurrencies, for example, instruct miners to disable sleep mode.
1 Like
Sleeping computer cannot mine. It will hurt your earnings, not the devices when they go to sleep.
1 Like
Btw what factors going to up price(snt) for space in final network?
I am thinking if we look on this as isolated idea if is then true that the more cost for 1mb in network then less snt I will get (but probably more value in fiat if I convert itā¦)
Understood. But what about the suspend/resume laptop pattern? A commonly occurring instance of network interruption.
So what can farmers expect for interruptions of eg: 1, 5, 30, 60 minutes, or longer such as 12, 24, 48 hours?
Keeping in mind that in these instances they come back online with all data and keys still in RAM. They might have the same public IP, or might have a different one.
5 Likes
HDDs do not like to be turned on and off.
Also temperature changes are important, solder joints do not like to expand and contract often.
3 Likes
I donāt think so.
As I said earlier, RAM and disk storage are not much different.
It is possible to store RAM contents on disk and possible to store disk contents in RAM.
Itās not a problem for example to create virtual machine on one physical machine, put it into suspend, transfer to second physical machine and restore its state there.
What outside observer can only see is disruption in network communication.
But if you treat network problems as a sign of restart, then there will be lots of false positives.
3 Likes
There are two most common failures on spinning discs, motor dies from too many on/off cycles or bearing dies from running 24/7.
Spinning HDDs are history anyway 
Temperature changes and solder joints - yes, lot of gaming GPUs die from this, they get very hot, but for other parts of PC it is very rare if cooling is ok.
1 Like
In the initial phase of SafeNetās operation, the number of redundant copies can be increased from, say, 7 to 10, and that should take care of the matter mathematically. Even with a geographically small area of network operation and the occurrence of some catastrophic event, the probability of losing all copies would be negligible, and once a certain global level of SN development is reached, one can return to the default assumption of maintaining 7 copies.
It is important, however, that when a computer is shut down, the data is not deleted, but merely put into hibernation, which could also provide an additional level of redundancy for the network (such a case, where a computer in Ghana was shut down due to a power outage, saved Maersk from disaster, as it was the only computer that was not affected by the Petya cyber-attack).
4 Likes
Beg to differ - there is still a huuuge installed user base - and anyway one of the projects principles was to make efficient use of otherwise under-utilised resources. I suspect it will be a long time (10 yrs +) until the last spinning platter dies.
As for the power on-off thing, ultimately its all down to temperature, embrittlement due to overtemp for a looooong time or repeated hot-cold shocks. In general Iād say if you want a PC to last a long time, switch it on and leave it on. If you want your gaming machine to last a long time, get better cooling.
1 Like
It was a little provocation, I was waiting who is going to protest first 
I know spinning HDDs will be with us for a long time, there are still people using CDs even floppy discs around. It is more about economics of using such devices.
I was doing some math lately, with current price of electricity, energy efficiency of new HW and cheap prices of some new HW, it doesnāt make sense to keep old machines running. Even when you spent zero on HW itself, electricity bill will catch you. Of course that is highly dependent on where you live, HW prices are similar around the world, electricity prices can vary a lot.
1 Like
My olā ENIAC, helps keep the house warm in the winter, so offsets the winter electric bill.
Nothing like a bowl of popcorn while huddled around the olā glow tubes to keep warm. 
1 Like
Plays havoc with the thermostat though.