It is a common perception that TCP does not work well over long-delay networks, particularly satellite. Although this is true in comparison to traditional T-1 or DSL lines, latency can be minimized to an extent where even most mission-critical applications can work perfectly over satellite.
Latency is defined as the amount of delay, measured in milliseconds, that occurs in a round-trip data transmission. Latency is unavoidable in any Internet connection and it can only be minimized using a better managed network. Not directly related to speed, latency can be an issue with all networks including satellites.
Latency is caused by several factors including the number of times the data is handled along the transmission path (by routers or servers for example). The GEO satellites used for two-way Internet service are located approximately 23,000 miles above the equator. This means that a round-trip transmission travels 23,000 miles to the satellite, 23,000 miles from the satellite to the remote site, and then as the TCP/IP acknowledgment is returned, another 46,000 miles on the return trip for a total round trip of over 90,000 miles. Depending on your latitude, this distance to the satellite could be even greater.
Each time a data packet ‘hops’ (i.e. handled by a device along the path) several milliseconds of latency are introduced. The physics involved account for approximately 550 milliseconds of latency, a limitation shared by all satellite providers. This latency can negatively affect the performance of many business applications and software that requires real-time user input.
In addition to transmission times, there are other factors that contribute to the total latency experienced by the end user. These include quality of internal network, IP/satellite translation overhead, speed of upstream connections, and most importantly, congestion of the satellite network itself. At VSAT Systems, we do not oversubscribe our networks and all our infrastructure is company owned and built from the ground-up exclusively for IP. Thus, our customers typically experience latency under 700 ms (and often under 600 ms) which is significantly lower than competitors who have latency as high as 900-1,500 ms.
Terminal emulators without local-echo (such as Citrix) can also be affected by latency, depending on the underlying application and configuration. If you have an application that is particularly sensitive to latency—such as online gaming—satellite may not work for you.
It is recommended that customers check with their software vendor to confirm how specific applications are affected by latency. Our sales engineers can give you an idea of what to expect after analyzing your requirements. For more questions contact us.