|
|
On Wed, 5 Oct 2005, Leonard Isham wrote:
On 10/5/05, Mathias Sundman <mathias@xxxxxxxxxx> wrote:
On Wed, 5 Oct 2005, Leonard Isham wrote:
On 10/5/05, Mathias Sundman <mathias@xxxxxxxxxx> wrote:
On Tue, 4 Oct 2005, Leonard Isham wrote:
Bad connectivity is bad connectivity a VPN will not change that.
If you application would fail with the connectivity loss it will with
the VPN, or without it... game over... unless you can resolve the
connectivity issue(s).
Not entirely true I believe. If "normal" connectivity is lost, some router
on the way usually return ICMP destination-unreachble packets causing the
TCP connection to be reset instantly. If you run your TCP application over
an OpenVPN tunnel packets will just be silently dropped during the network
outage until OpenVPN detect the network is down via --ping-restart, so TCP
sessions inside the tunnel will survive until they time-out.
Depends on quite a number of "things," including traffic shaping that
silently drops packets.
What I was attempting to explain is that if the loss of connectivity
is long enough to interrupt the application a VPN will not fix the
application. I have seen quite a few applications designed to work on
a LAN that are pushed into a WAN and or VPN environment and they
simply do not scale to the expanded environment.
Also frequent connectivity losses can really mess up connections.
Sure VPNs can make things worse under some circumstances. What I was
trying to say was just that it can also help, as he was asking for under
some circumstances.
A live example:
I have a bad wireless (802.11B) network at home that causes my laptop to
loose network connectivity from time to time. All my sessions running
directly over internet then dies, while my SSH session to my email server
that runs over an OpenVPN link survives as my TAP-Win32 adapter never
reports "down".
I understand your frame of reference now.
My frame of reference is the connectivity loss is not local to either
system and would not cause an interface to flap (locally that is). In
an internet environment, which I believe the OP is in the likelyhood
if local interface flap is usually minimal.
Yes, but my last post was just an example of simple scenario where OpenVPN
can help with a network problem. The problem does not have to be local.
Say that a router on the way has a WAN link that goes up and down. This
router will then return ICMP error messages to any host trying to route
anything over this link and thous causing applications to break. With an
OpenVPN link running over this, these ICMP error messages will never reach
that host, and thous applications has a better chanse of surviving a short
network outage.
If the problem is only with silently dropped packets across the network,
which might be the most probable cause, then I agree OpenVPN will not make
things any better than the link is from the beginning.
I'm not saying that OpenVPN will solve all the OP's problems, just that it
might help with some local and non-local network problems as opposit to
your original post (quoted at top) where you say that "the game is over"
as soon as you have connectivity problems. That simply depends on exactly
what kind of connectivity problems he is having.
--
_____________________________________________________________
Mathias Sundman (^) ASCII Ribbon Campaign
OpenVPN GUI for Windows X NO HTML/RTF in e-mail
http://openvpn.se/ / \ NO Word docs in e-mail
____________________________________________
Openvpn-users mailing list
Openvpn-users@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/openvpn-users
|