|
View:
New views
6 Messages
—
Rating Filter:
Alert me
|
|
|
Slow speedsHello,
I've just setup my VPN on a remote CentOS box and I connect to it via the Windows XP network connections feature. My internet connection is 2Mbps and I can get 2Mbps http downloads from the VPN server without any problems when I'm accessing it from the outside world (via my ISP gateway). However, when I do the same thing inside the VPN download speeds are steady at 55 KB/s or roughly about 512 Kbps which is 4 times lower than my actual link between me and server. I'm not sure whether the VPN is expected to behave like this? Any suggestions? Generally, logs do not say anything is wrong and everything is going smoothly - CPU utilisation and Memory utilisation are all fine. Thanks, -V. ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Poptop-server mailing list Poptop-server@... https://lists.sourceforge.net/lists/listinfo/poptop-server |
|
|
Re: Slow speedsIf you have any measurable packet loss over the path, this is amplified
by the VPN by nature of how it works. Have you tried OpenVPN? It uses UDP encapsulation rather than the PPP over GRE that is used by PPTP. Consider MTU values. If the MTU of the underlying link is the same size as the MTU of the tunnel, then every tunnel packet which is full size may require two packets of the underlying link. -- James Cameron http://quozl.linux.org.au/ ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Poptop-server mailing list Poptop-server@... https://lists.sourceforge.net/lists/listinfo/poptop-server |
|
|
Re: Slow speedsHi James,
Very strange, but OpenVPN gives me the exact same result. Only about %25 of the connection is utilised and I tried changing ports just to be certain that my ISP is not traffic-shaping. What might be causing this? btw, there's no packet loss between me and the VPN server. -V. On Wed, Sep 9, 2009 at 4:13 AM, James Cameron<quozl@...> wrote: > If you have any measurable packet loss over the path, this is amplified > by the VPN by nature of how it works. > > Have you tried OpenVPN? It uses UDP encapsulation rather than the PPP > over GRE that is used by PPTP. > > Consider MTU values. If the MTU of the underlying link is the same size > as the MTU of the tunnel, then every tunnel packet which is full size > may require two packets of the underlying link. > > -- > James Cameron > http://quozl.linux.org.au/ > > ------------------------------------------------------------------------------ > Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day > trial. Simplify your report design, integration and deployment - and focus on > what you do best, core application coding. Discover what's new with > Crystal Reports now. http://p.sf.net/sfu/bobj-july > _______________________________________________ > Poptop-server mailing list > Poptop-server@... > https://lists.sourceforge.net/lists/listinfo/poptop-server > ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Poptop-server mailing list Poptop-server@... https://lists.sourceforge.net/lists/listinfo/poptop-server |
|
|
Re: Slow speedsOn Wed, Sep 09, 2009 at 04:42:29AM +0100, Vanja Milosevski wrote:
> Very strange, but OpenVPN gives me the exact same result. Only about > %25 of the connection is utilised and I tried changing ports just to > be certain that my ISP is not traffic-shaping. > > What might be causing this? Many possible causes, too many to enumerate. Packet loss or latency are the most likely. > btw, there's no packet loss between me and the VPN server. How have you measured this? If you have measured using ping outside the tunnel, then perhaps the loss is specific to the tunnel protocols. If you have measured using ping inside the tunnel, this might be meaningless. Look for TCP retransmissions or lost TCP segments using wireshark on the network interface of one of the endpoints of the tunnel. Do that by capturing packets while a slowed transfer is happening, then select that TCP stream in wireshark and look at the packets individually. wireshark highlights retransmissions or lost segments in certain ways ... I can't recall exactly how. But compare it to a known good TCP connection that isn't affected. -- James Cameron http://quozl.linux.org.au/ ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Poptop-server mailing list Poptop-server@... https://lists.sourceforge.net/lists/listinfo/poptop-server |
|
|
Re: Slow speedsHi again,
Nothing interesting in Wireshark so far. Out of curiosity, how much overhead would a VPN ideally introduce in terms of throughput? -V. On Wed, Sep 9, 2009 at 4:59 AM, James Cameron<quozl@...> wrote: > On Wed, Sep 09, 2009 at 04:42:29AM +0100, Vanja Milosevski wrote: >> Very strange, but OpenVPN gives me the exact same result. Only about >> %25 of the connection is utilised and I tried changing ports just to >> be certain that my ISP is not traffic-shaping. >> >> What might be causing this? > > Many possible causes, too many to enumerate. Packet loss or latency are > the most likely. > >> btw, there's no packet loss between me and the VPN server. > > How have you measured this? If you have measured using ping outside the > tunnel, then perhaps the loss is specific to the tunnel protocols. If > you have measured using ping inside the tunnel, this might be > meaningless. > > Look for TCP retransmissions or lost TCP segments using wireshark on the > network interface of one of the endpoints of the tunnel. Do that by > capturing packets while a slowed transfer is happening, then select that > TCP stream in wireshark and look at the packets individually. wireshark > highlights retransmissions or lost segments in certain ways ... I can't > recall exactly how. But compare it to a known good TCP connection that > isn't affected. > > -- > James Cameron > http://quozl.linux.org.au/ > ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Poptop-server mailing list Poptop-server@... https://lists.sourceforge.net/lists/listinfo/poptop-server |
|
|
Re: Slow speedsOn Wed, Sep 09, 2009 at 05:50:17AM +0100, Vanja Milosevski wrote:
> Nothing interesting in Wireshark so far. Okay. I've no other suggestions. > Out of curiosity, how much overhead would a VPN ideally introduce in > terms of throughput? Ideally, nothing at all. We don't use ideal networks though. Every IP packet delivered to the tunnel entrance has to be encrypted and encapsulated in another IP packet. The encryption adds a small amount of extra bytes. The encapsulation for PPTP uses PPP, and this increases the packet size slightly. Once a packet does not fit, it is split for transmission as two separate packets. The additional packet has a header which is a new cost compared to the original packet. In my experience, there usually is an overhead, but I don't have figures to hand on what the overhead is. -- James Cameron http://quozl.linux.org.au/ ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Poptop-server mailing list Poptop-server@... https://lists.sourceforge.net/lists/listinfo/poptop-server |
| Free embeddable forum powered by Nabble | Forum Help |