dotgnu-general
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [DotGNU]Re: MOSIX as WOS


From: David Nicol
Subject: Re: [DotGNU]Re: MOSIX as WOS
Date: Wed, 17 Oct 2001 14:07:34 -0500

Bill Lance wrote:

> This is interesting. I did some experimenting with
> MOSIX some time ago and, if I recall, there was a
> stronge emphasis on using a totally secure network.
> MOSIX left it's nodes open naked to eachother with no
> internode security measures in place at all.
> 
> How can you use this safely across the internet?

recent MOSIXes refuse mosix peer connections
from IP addresses that are not listed in their configuration files.

Additional security, such as a whole lot of one-time-only auth
cookies to prevent connection hijacking (but not man-in-the-middle
attacks and other science fiction) would not be that hard to add,
but I would try to be objective about both the risks involved and
my medication levels before embarking on that adjustment: you need
to intercept as well as originate to carry out a successful TCP 
hijacking as I understand it, and that requires hardware access to
an internal wire, or compromise of an internal system, to get,
at least using PPPoE style DSL connections.  Cable modems, where
it is very slightly possible that you as a person who is reading this
are not the most haxorly person on your coax segment, are a slightly
different story.  The standard mechanism against TCP connection
hijacking applies, if you really are seriously worried about it, just like
telnetting or receiving e-mail, and that is, secure tunnels.  You set up
a secure virtual network and then peer over it.  Yes the lag goes up.

The question becomes, is the gain from MOSIX peering worth the
trouble of configuring and then using a secure tunnel, in both CPU load
and bandwidth use?

MOSIX 1.4.1-Pre2 for Linux 2.4.12 (released monday) is the latest
version.  Inter-node compatibility is indicated by the first two
version numbers.  The third version number is patch level within
a protocol.  There are also some configurable parameters that have
to match in order for peering to occur.

-- 
                                           David Nicol 816.235.1187
                                            1,3,7-trimethylxanthine


reply via email to

[Prev in Thread] Current Thread [Next in Thread]