Troubleshooting asymmetric throughput

[FloydTheDuck]

We have a WAN that is setup with BGP, DMVPN, and EIGRP. 10 sites all connect to the service provider via BGP and to each other via DMVPN with EIGRP running on top. We're experiencing an issue where any egress traffic from Site A to any other Site is slow (1-2Mbps, where 40+ is expected).

Site A -> Site B or C --- 1-2Mbps

Site B -> Site A or C --- 40Mbps

Site C -> Site A or B --- 40Mbps

The router at Site A is a 2951. The DMVPN shares the same physical circuit as our Internet, which running a speedtest gets 40/35 (down/up).

The edge connection is a Cisco ASA 5508 HA pair.

Any suggestions on items to check?

[Larry Sullivan]

Hi,

First and foremost, how are you testing throughput?

Does site A public IP to site B/C public IPs have low throughput or just through DMVPN?

[FloydTheDuck]

Testing with iperf between servers that are connected off the embedded switch in the router at each site (the router runs BGP and DMVPN for the site).

Internet speeds at each site seem fine. We tested with 6 different sites and Site A seems to be the common denominator, only having reduced speeds over the DMVPN only with egress traffic from Site A.

I'm beginning to suspect it may be an issue with the ISP, but I don't know a way to pinpoint that (or to prove that there's nothing on my end, at least).

[Larry Sullivan]

Yeah, that's a tough one.  If you are so inclined and have the time and it's not service impacting, you could build a basic GRE between site A and B or C and try running the iPerf test through that.  Will give you an indication if it's the actual path or DMVPN/EIGRP related.