I've rented a machine with 24GB ram and 8 cores. Matthew has the login info. Hopefully others can donate more testing resources.
On Fri, May 3, 2013 at 8:46 AM, Matthew Toseland <t...@amphibian.dyndns.org>wrote: > On Friday 03 May 2013 13:31:04 Michael Grube wrote: > > > > > > Also, the reality may be even worse: Even without attackers, churn in > the > > > network causes "natural" degeneration; we randomise periodically to > try to > > > correct this. > > > > > > Note that we are probably immune to their third attack due to using a > > > commit/reveal on swapping. However the first two are real enough and > can't > > > be beaten by commit/reveal. > > > > > > We could implement this tomorrow. What exactly do we need to determine > > > before doing so? We could validate the implementation, and compare it > to > > > current swapping, in a many-nodes-one-VM simulator (we can simulate > approx > > > 100 to 500 "real" nodes on one high end system). > > > > > > > That's interesting. What do you consider high end? If it will actually > > help, I'll rent some cloud hardware. I'm not sure about supporting 10,000 > > nodes, but maybe 2 or 3 thousand from me. > > Relatively high CPU and RAM. It's only needed because we still encrypt, of > course. We could make this much more efficient by modifying the node a bit. > > Obviously we slow things down a lot, run relatively few requests, small > stores etc. > > See freenet/node/simulator/RealNode*Test > > Note that I've never distributed this between multiple systems... > > It's a spectacularly inefficient simulation, but since it uses the real > code, it's useful for sanity checking. Might be worth making it more > efficient in future... >
_______________________________________________ Devl mailing list Devl@freenetproject.org https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
