|
Lets not go on a MojoHost / Brad bash fest here.
Sam was correct in stating that it isn't the end-all of network performance. We utilize GNAX's network as one of our providers and their network is 'enhanced' by the RouteScience platform. While the device itself can dramatically improve your outbound routes it has no effect what so ever on your in bound paths.
In the traceroutes as posted above it would be a relatively safe assumption to say that the main problem isn't so much the lack of the FCP making a difference on the return path but the pain of the inbound path.
With a relatively limited mix of carriers with significant US presence on their networks all of that must be peered off at some point which adds to the undesireable paths to reach peering points for the handoff.
BTN has a very weak east coast network, Telia very weak west cost, Seabone / TelcomItalia is very poor in the US in general which leaves only Level3 and GlobalCrossing amd Cogent for solid US coverage.
The only way to solve the inbound path issue is to have direct connectivity with as wide of a carrier base as possible to contain the traffic to within that carriers network which *should* but not always be more optimal of an inbound path (sprint, at&t, mci/verizon, savvis, etc).
For all it's glory the NAP of the americas has a big disadvantage. It's in Miami. While thats great for connectivity to the southern hemisphere and parts of europe it's not advantageous for american destinations. As you noticed there is very little peering if any going on in miami in all those traces things were backhauled to another city further north, ie washington to make the handoff and then back down to miami.
|