In January our company solicited bids from several T1 providers in the St. Louis area. The initial list included providers we already knew and some from a general internet search for alternative providers: AT&T, Sprint, XO Communications, MegaPath, and Covad. After our preliminary bids from the companies we narrowed the choice to XO Communications and MegaPath.
XO Communications was known to us from other companies who used their service in our area. They were very professional but their price was higher than MegaPath's and they were unable (or unwilling) to guarantee our 3rd party hosted VoIP telephony traffic which was paramount to our decision.
MegaPath on the other hand was less expensive, claimed to have the same level of local support staff as XO Communications and guaranteed they could handle our 3rd party hosted VoIP telephony traffic by implementing QoS/CoS. After many conference calls with both providers and weeks of due diligence we finally selected MegaPath.
The install of the bonded T1 (3M up/down) went smoothly and was completed in March. Unfortunately several issues surfaced immediately.
1. The first hop ping time was 70ms and average ping times to major sites (e.G., google.com) were over 120ms even despite less than 2% utilization on our line. This made general web surfing noticeably slow and prevented our VoIP service from working at all.
2. Our geocoded location was now showing Washington D.C.
Considering we're in St. Louis and use geocoding to route shipments this was problematic.
3. There was no QoS/CoS for our VoIP traffic on the line as promised.
Problem 1 was actually due to the fact the salesperson lied about having "local" service.in reality they contract for a local loop to their "long haul" connection somewhere in St. Louis. Ironically the local loop provider is XO Communications. The long haul then provided a frame relay style service to their actual network in Washington D.C.
This was why our first hop had so much latency, it was actually a VPN-like hop over a very long distance through who knows how many actual routers. It also explained Problem 2 - our public IPs were assigned to their Washington, D.C.network. We ultimately fixed Problem 2 by hard coding the geolocation in our database. This has worked short term but the minute we open a second distribution center things will break again.
As for Problem 1, we opened trouble tickets with MegaPath. They claimed our first hop was within their SLA of
0 comments