How does a point to point T1 work?

Discussion in 'Networking & Security' started by vapb400, Jul 9, 2009.

  1. vapb400

    vapb400 Gawd

    Sep 19, 2004

    I have a client who has multiple offices and currently everything is tied together via IPSec tunnels via ASA5505s main office has 3/3 bonded T1 and the branch offices have T1 or DSL

    The latency is too high to get good quality with VoIP and I was thinking about the possibility of doing point to point T1s.

    Can anyone explain to me how that works? I have literally no experience with Point to Point T1s and I have never seen them used.

    So if I have my external IP of or something and my internal DHCP scope ofof
    Where would I fit that in?
    Could I route it so the workstations in the branch offices get DHCP from the Server in the main office? Or how would it be done?

    We want to use VoIP phones and we would be running Thin Clients on the other end. The only thing is that there are network printers on the branch-office side as well so I need them to be seen from the main servers easily.

    Its about 50 computers in the head office and 10 thin clients in the remote office + 2 network printers

    I hope that made sense and I hope someone can shed some light on it for me!

    Next step is to look into MPLS, but I was wondering if someone could explain this for me first.
    Last edited: Jul 9, 2009
  2. xphil3

    xphil3 [H]ard|Gawd

    Nov 11, 2005
    What VoIP solution are you working with? What codecs(G.711, G.729)? How high is your latency? Ive seen 100+ ms calls where call quality is not degraded but you do notice some hefty delays. Jitter is the VoIP killer that you have to worry about. Wes, feel free to chime in :D

    Point to point circuits are just that, a virtual circuit(PVC) between one device at one location and another device at another location(generally geographically despersed). You generally lease these lines from a provider and they are local to YOU only. No internet and since they are local to YOU only they can/usually are private range addresses. This is probably the most basic description I can give you.
    Not sure what you mean by where would that fit in? but yes, you can run a DHCP server from the main office and have branch offices acquire their addresses from that server. Look into DHCP relay(for the ASA) or ip-helper addresses for routing. Use the google box.

    Also, since you're asking about P2P circuits I can assume that you probably aren't doing any kind of QoS? Implment that immeadiatley. I have run around 10 simultaneous calls from on cable line(IPsec'd and GRE'd) without any degredation in the call. At the time, upload was less than 1Mbit. Also, keep in mind your Codecs mentioned above.
  3. RavinDJ

    RavinDJ 2[H]4U

    Apr 9, 2002
    I have a client who has a point-to-point running... there are two (2) Adtran routers (one on each end) and one location is and the other is

    The Adtran router on one end is and the other location is

    It actually works pretty good and the PTP works pretty much 99.999% of the time.

    Hope this helps :p
  4. Protoform-X

    Protoform-X [H]ard|Gawd

    Jan 30, 2002
    You'll be amazed how much faster the ThinClients will work over a low latency PTP connection.

    Here's how we typically setup our PTPs:
    - --> Main office PTP router LAN(server on this subnet typically handles main office DHCP)
    - --> Main office PTP router WAN
    -In between --> Service provider equipment... Usually an Adtran on both sides; T1 circuit.
    - --> Remote office PTP router WAN
    - --> Remote office PTP router LAN(DHCP for office on this interface so that the office can operate without a connection to the main office)

    I also like to setup a VPN failover on the PTP router. You can set a floating static route so if your PTP route goes down you automatically start pushing traffic out to your firewall which has a VPN connection to the remote office. Just remember that you'll need more than one ethernet interface on your PTP router to make this work(eg... no Cisco 1700's).

    I prefer to have a DHCP server at each remote location just in case the main office circuits go down or power goes out. It helps the remote office be a little bit more self contained in system down situations. Of course, you can always use the ip helper-address command to unicast your DHCP information from your main office.

    Let me know if you need more help with this. I have a ton of clients about the same size as you're describing using PTPs... Maybe my experience can help you.