I ran a trace route from my company to magicvortex.com. The trace took a total of 18 hops. I am sure that depending on the ISP a user is attached to the internet through, the hops would vary greatly both in number and routers hit. My question is: At any given router that the "large files" pass through, how can you guarantee that there is no limit set that would block the transmission of these files? Are there circumstances under which transmissions will not be successful? Do you automatically seek new routes in the event of a failed transmission?
The architecture of the magicVORTEX transfer service focuses on using simple and standard-based technologies as the foundation of an extremely reliable file transfer system. When files are being transferred, they are broken into small packets that range from 1KB to 16KB (the size is user configurable). These packets are then streamed sequentially until all the packets have been sent. The Vortex Relay, which is at magicvortex.com, is responsible for tracking the progress of which packets have been transmitted successfully. If any packets fail to reach their intended recipient for any reason, they will be sent again. This re-transmission will continue until either the file is completely sent or the file has expired based on a predetermined file expiration date.
Any router encountered, between the source machine and the Vortex Relay, would only see a 16KB maximum HTTP request. Each packet is an independent request (i.e. the network never perceives a "large file" as a single entity).
One possible way in which a file transfer could be prevented would be if you were operating behind a firewall which explicitly blocks either "magicvortex.com" or the IP address it is mapped to. Otherwise, firewalls do not typically pose any problem.
By utilizing standard protocols, redundancies within our service provider's network greatly improve the reliability of the magicVORTEX service. Our ISP's network dynamically determines the most efficient route by monitoring the health and performance of each subnet.