The default mechanism for connecting from on-premises infrastructure to cloud data centers is via IP networks, which provide the communication fabric of our massively interconnected world. Traditional methods of sending and sharing data over IP networks rely on a networking protocol known as Transport Control Protocol (TCP). TCP, and protocols such as Hyper Text Transfer Protocol (HTTP) and File Transfer Protocol (FTP) that are built on it, works well for sending relatively small data sets over the Internet.
However, the reliability and congestion control mechanisms of TCP do not deal well with high latency and packet loss on long-distance networks, thereby causing transfers to become impractically slow. Unless the cloud data center is very close from a network perspective, a standard TCP stream does not provide sufficient throughput for large-scale cloud upload tasks.
A common approach to solving the throughput problem, one based on widespread misconceptions, is to purchase more bandwidth. If the goal is to help a large number of users achieve higher aggregate throughput, some improvement may be realized. But a bigger pipe does not address the underlying limitation of TCP that results in extremely inefficient utilization of bandwidth in high-latency scenarios, a problem that becomes worse in the presence of packet loss.
In fact, more bandwidth makes literally no difference in many practical situations: above a minimum bandwidth-latency product threshold, TCP throughput is a function of latency and loss and is completely independent of bandwidth. So it is quite likely that the transfer time for a given file will be exactly the same over the higher bandwidth connection.
Simply buying more bandwidth can be the very definition of throwing good money after bad.
The use of IP networks to reach the cloud is a given. Whether public Internet or private connections, that’s how the world is connected today. The challenge is to figure out how to use these networks in a more optimized manner so they can better handle the massive data load of the cloud — which is what Signiant Flight is designed to do.
By minimizing the impact of latency and loss so the entire bandwidth of the connection can be utilized, Signiant’s proprietary transport technology takes the network bottleneck out of the equation.