Link to home
Start Free TrialLog in
Avatar of ssuser1
ssuser1

asked on

pt to pt connection very slow

I have a pt to pt connection between two sites running a T1.  While pinging and browsing each others network is fast.  Accessing shares or doing installations or copy information from site A to Site B is very slow
Avatar of rfc1180
rfc1180
Flag of United States of America image

More than likely a congested T1 (Saturated), even an issue with the T1 taking errors causing packet loss. What type of routers do you have connecting to the T1?
ASKER CERTIFIED SOLUTION
Avatar of Scott Gorcester
Scott Gorcester
Flag of United States of America image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of ssuser1
ssuser1

ASKER

We have cisco routers on both end. Even if only one machine is trying to access the other site it is slow and the router will show high utilization even with one user.  If the T1 is just between the two ports shouldn't it be enough.  
please report the output of the show serial interfaces

show serial int0/0 (replce 0/0 with your interface numbers)
Provide both ends.

Billy
Also, try an MTU ping test:

http://help.expedient.com/broadband/mtu_ping_test.shtml

Standard MTU sizes are 1500 bits. By tunneling through a VPN, you are adding 24 bits to the packets.

A typical ping is only a few bits, and a browsing of each others sites is only a few packets, enough to not notice smaller sizes.

I think by changing the MTU size of your tunneling routers, it will allow your typical 1500 bit packes to go through freely without fragmentation.

This problem is called Maximum Segment Size Exceded. By showing serial interfaces, as mentioned above, it should tell what the MTU size is on that interface.
Well, the MTU size is typically 1500 Bytes, and the VPN overhead is variable depending on the mode and encryption being used; If using a GRE tunnel, yes, the overhead would be 24 bytes. At any rate, this is typically not an issue for TCP hosts as they utilize PMTUD which is a mechanism that (Path MTU Discovery), as the name suggests, allows a host or network device to dynamically discover the lowest MTU along a path to a destination. This of course assumes that the host is setting the DF Bit, no firewalls blocking ICMP, and the host has the ability to reduce the size of data to meet the requirements end to end.
That's why I wanted to see ping tests, to see if ICMP echo is disabled between adapters. I have seen drastically reduced efficiency from an MSS exceded problem. I have also seen timed out web pages. So, with the browser not having problems, maybe things are good with the MTU sizes on the tunneling routers.
Agreed, something to look at.

Billy
There is an inherant delay in connection over the Internet that is not there on a LAN. Trying to use NetBIOS or SMB (version 1) over a VPN or other Internet-based conenction medium will result in VERY poor performance (usually unacceptably poor).

The "fix" is to use SMB version 2. It is ONLY available on:
  - Windows Vista, 7, and Server 2008
  - Samba 3.5 and above (including the recent releases of Samba 4 beta)

You'll be VERY hard pressed to find any NAS or embedded CIFS devices capable of running SMB version 2, so you're looking at a Windows 2008 or Samba 3.5 server to provide file services -- but again, the CLIENTS need to be one of the above OSes to make use of SMB 2's upgrades!

Read more here:
http://wikipedia.org/wiki/Server_Message_Block

Best Regards,

Dan
IT4SOHO