On Mon, 02 Sep 2013 16:50:11 -0700, Patrick Elliott wrote:

>> Sounds like ISP interference to me rather than the OS, though.
>>
> But, its not. There are a fair number of servers that, for security
> reasons, may disable the control packets, including routers in the
> primary backbones of the internet (or alternate paths). This means that
> tracing a problem, even in your own network, never mind someone else's,
> either nearly, or totally, impossible, using the normal methods.  The
> TCP/IP solution was specifically developed to a) do the same thing if
> you can't/don't want to, disable the blocks on those functions, b) get
> around issues, such as alternate routing, where you are still blocked
> from access, but you can't work out why, etc. Control packets are not
> "necessary" for normal operation of a network. Its not unknown, since
> there are some things you can do with them, other than route tracing,
> and pings, for them to be disabled, but.. its like closing a port, in a
> sense, if you can't talk to past what ever is blocking it, short of
> having, say, some way to proxy it, you can't use it at all, any more
> than you can talk to the port that has been closed.
> 
> So, yeah, its definitely the ISP's fault, in a sense, but.. again, this
> is just two commands that I am talking about. There are entirely test
> tools that rely on the ability to do semi-abnormal things, to get
> various kinds of information, not just from the internet in general, but
> just from your own network, and literally the **entire** toolset is
> broken, beyond use, because the sloppy "protection" in Windows can't
> tell legit testing suites from actually invalid traffic, and just
> handles them all, arbitrarily as though they are threats.

It doesn't really seem reasonable to me to hold the OS producer 
responsible for not being able to do - as you call it - "semi-abnormal 
things".

If the filter is behavioural, then of course it can't tell the difference 
between a diagnostic tool and actual malicious traffic.  Tell me, how 
would *you* code the software to tell the difference between 
"legitimately" wonky behaviour, and actually malicious behaviour?

Jim

Post a reply to this message