Help - Dropped packets across all Speedfusion tunnels


#1

I am begining to roll out 100 HD2 units with a head end 1350. The 1350 has three WAN connections, two internet and one MPLS ethernet handoff. The remote HD2s have a single MPLS ethernet handoff and a cable/dsl internet connection. Each location has three local subnets. We are redistributing 0.0.0.0 from the head end, all remote sites have a default outbound policy pointing all traffic to the VPN, there is one other policy pointing Vlan #3 (the guest VLAN) directly to the intenet connection. All sites have speedfusion tunnel back to the head end.

In all cases I am getting packet drops when I ping from the switch directly behind the head end to the remote end over the tunnel. I am not getting drops when I ping accross the MPLS connection directly to the HD2s MPLS connection, only when I go across the tunnel. I have attempted to disable cable optimization, I have enabled WAN smoothing in every way all the way up to MAX, I have taken one connection or the other out of the tunnel and I still get drops when I ping accross the tunnels.


#2
  1. May I know where you ping to (from the switch in head end)?

  2. Please try to ping with other devices (in head end and remote side). Do you see the packet loss?

  3. As you mentioned, there is no packet loss when you ping via the MPLS link. Are you using the same devices which I asked in the question 1 above?

  4. Beside the ping loss, have you facing any problem of your application over SpeedFusion tunnel?

Fyi, this is not the apple to apple comparison. From the ISP point of view:-

  • ISP sees the ICMP packet when you ping via MPLS link.
  • ISP sees the UDP 4500 packet when you ping via SpeedFusion tunnel.

ICMP was optimized by most of the ISP. Hence, you will get a good result normally. I suggest getting help from the local Peplink partner (your purchasing point).

Thanks.


#3
  1. yes from the switch in head end or from a workstation either way drops same amount of packets
  2. yes I see around the same amount of packet loss pinging other devices beyond the gateway. I also see packet loss both directions (from switches at remote site to core, and from core switches to remote switches)
  3. yes there is only one 1350 at the head end
  4. yes users are complaining about getting dropped in Citrix and other applications that require a persistant connection.
  5. I am reaching out to my partner for assistance as well :wink:

#4

Just for note: The fix for this issue was configuring the advanced options in the Speedfusion setup to MAP MPLS connections to MPLS connection and Inet connections to Inet connections. Since configuring this option in the tunnels the packet loss has gone completely away at all sites. This “advanced settings” option can be found when configuring the speedfusion tunnel in IC2 at the bottom left last page of a point to point tunnel or in link configuration of a star tunnel.