>* Malkus Lindroos <malkus at iki.fi> [2007-08-21 15:51:51]: >> Matthew Toseland wrote: >> > On Tuesday 21 August 2007 00:13, Matthew Toseland wrote: >> >> On Monday 20 August 2007 21:59, Malkus Lindroos wrote: >> >>> Of course it depends on what the ping time is supposed to reflect? At >> >>> least for my nodes it seems to reflect the load of the network the no= >de >> >>> is connected to, not the load of the node or its internet connection. >> >> It's a combination of the round trip time under ideal conditions (<0.2= >ms on=20 >> > my=20 >> >> LAN), and the effect of the traffic on the links between the nodes=20 >> > (including=20 >> >> rate limiting). >> >=20 >> > Sorry, not just network load, ping time is highly sensitive to CPU load= > on=20 >> > both ends too. This is why it's such a useful metric; we can't easily= >=20 >> > directly measure CPU load, and even if we create platform specific hack= >s,=20 >> > we'd be fooled by idle-time processes etc. >>=20 >> OK - so there comes the problem. My node CPU load is only about 5-10%. >> It has clearly idle time, and the maxpingtime is the primary limiter. >> Some connected nodes have lower ping times than others - while many >> nodes have ping times of 3000-4000 ms, some stay at 500. My node also >> has plenty of free bandwidth and its hard disk is not overloaded. >>=20 >> So in effect, the average ping is reflecting the CPU loads, bandwiths, >> disk usage, etc. of my node + the nodes is connected to. This eliminates >> load distribution among the nodes, as my node does not take more work >> from the overloaded nodes because the overloaded nodes are limiting it. >> When the load of the neighbouring nodes worsens, they also limit my >> nodes ability to take load. I.e. the main problem in the ping times is >> that it is determined not only by my node, but also the nodes that it is >> connected to. In addition, the ping time is determined by a principle of >> the weakest link - the ping time of a node becomes high if only one of >> the nodes is overloaded. >>=20 >> This also provides a possibility for a denial of service attack against >> the 0.7 network - from the code it is not too difficult for someone >> wishing to impede free speech (or the 0.7 against 0.5 networks) to >> create bogus nodes that report way too high ping times. This would >> severely hurt the 0.7 network from reaching its potential.
>Using the median insteed of the mean solves that problem... Moreover >really high ping times are not possible because there is a timeout. If above assumtions are correct, one could say that: - in every ping time the local node has to a remote peer not only the local load characteristica (cpu/mem/hd load) but also the remote load is incorporated and the network between them: - - local load because of local work like cpu-using programs (parallel runing boinc, etc) - - remote load because of all the above - - connection load like network congestion due to ISP/LAN traffic or bandwidth Also we have different setups: - a) two nodes running on the same computer - - here local and remote load are effectively the same so a "doubled load" on one computer and none on the other - b) two nodes running on distinct computers but over a very highspeed network (=LAN) - c)two nodes running on distinct computers and around half the globe (normal usage) let's draw these up: _________ |___a)___ |___b)___ |___c)___ local load |2x |1x |1x remote load |none |1x |1x network load |none |small |1x As the average ping time is used to determine the local load - the network load is irrelevant - the remote load is irrelevant Therefore the local load can safely be calculated from the _smalles ping time of all peers_ and not an average of these because - for a) the determined local load actually would be doubled because on one computer there are two nodes running that affect each other with their own generated load - - irrelevant as this would not create a lower ping time as the normal scenario c) - for b) the ping time can be used as b) incorporates the local load (+ extra load from remote but not network) - - the lowest ping time would account for local and remote load - for c) the ping time can be used as c) incorporates the local load (+ extra load from remote and also the network) - - the lowest ping time would account for local and remote load and also the network load, so the ping times are actually higher than they need to be as the network has a large impact to the ping time that's understood as "local load" Am I way off here? :) >>=20 >> Because of this, there should at least be an option to disable the use >> of maxpingtimes in the config for now and see if the network would >> become faster. >>=20 >Again, it's not a problem unless most of your peers are missbehaving... >And as you choose them it's up to you. Not in Opennet >NextGen$ >--OBd5C1Lgu00Gd/Tn >Content-Type: application/pgp-signature; name="signature.asc" >Content-Description: Digital signature >Content-Disposition: inline >-----BEGIN PGP SIGNATURE----- >Version: GnuPG v1.4.6 (GNU/Linux) >iD8DBQFGyuFGU/Z/dHFfxtcRAuPCAJ9GiPxLXwHi3QLfgj/U1uqBpXGh0gCbBtHR >Cc7f3JVUSjzRx70t04zWKuQ= >=/QCN >-----END PGP SIGNATURE----- >--OBd5C1Lgu00Gd/Tn-- >--===============1326198163== >Content-Type: text/plain; charset="us-ascii" >MIME-Version: 1.0 >Content-Transfer-Encoding: 7bit >Content-Disposition: inline >_______________________________________________ >Tech mailing list >Tech at freenetproject.org >http://emu.freenetproject.org/cgi-bin/mailman/listinfo/tech >--===============1326198163==--
