According to the Systemdocs of the Self encryption:
As data is saved to a user's virtual hard-drive, it is broken up into a minimum of three chunks, hashed and then encrypted. To further obfuscate the data, every chunk is passed through an XOR function using the hashes of other chunks.
Each chunk is then broken in thirty two pieces and key value pairs are added to a table on the users computer called a data map
So we have four step:
2,/Aes256 encryption with the hash of next chunk.
3./XOR encryption with the next encrypted chunk.
4./Generate 32 new pieces (of each twice encypted chunk) with the data dispersal algorithm
What the size of the chunks after the four step?
In relation to the old system, without data dispersal, the Datamap increases greatly.
Can we have problems with slow boot as long as we should read the Datamap to access other files?
With the data dispersal, especially with no large files, do not generate a lot of work to Managers?
it is chunk/groupsize (1Mb / 32)
This does not affect data map, it’s handled in lower layers to transport the data.
Not really as they process the files anyway so instead of 32 * 1Mb payload they have 32 * 1/~32 payload to deal with.
We have not implemented the dispersal yet as we removed it to test another approach first (IDA modules written though, so simple to test). We may have a mechnism where you do not directly connect to managers (when I say may I mean have ) and this has further security implications, nobody will know where you connect to on the network so if they wait in your close group you may never be there
Ah this part is incorrect, I see the issue you were having, this does not affect the datamap. Sorry for the confusion there.
You’re full of surprises. I may have to update my other thread SAFE Network Infrastructure (Security).
If you’re saying what I think you’re saying? We are able to mitigate the 32 Node connection link in the chain. I don’t have any details to explain it, so I’ll just let you reveal it when you’re ready.
OK, thanks. I am translating the Systemdoc to Spanish and was a little puzzled with this part.
Many thanks for picking it up
Apologies @digipl, that’s my fault. I’ll get this part corrected.
@dirvine is this the thing where the close nodes are used as a “proxy group”? And can you explain a little more about it? The dispersal will be off the table for Testnet3?