Delays and Dilemmas
As you watch evening news reports from journalists located around the globe, you often can detect a split-second delay before the field reporters respond to questions from the anchors. A similar scenario may also occur when retrieving information from your company’s database, which could be hundreds or even thousands of miles from your office or hotel room.
Such delays happen frequently when communications rely on satellite, but in most cases, the delay, known as satellite latency, is barely noticeable. Satellite latency is the time delay that occurs when data or voice communications are transmitted via satellite. The signal travels from the ground to the satellite, which is orbiting about 23,000 miles above the equator, and then returns to Earth. The signal’s round trip — often referred to as a single hop — takes about 250 milliseconds, or a quarter of a second.
While engineers have been dealing with satellite latency issues for 30 years, the average person usually is not aware of this delay until they experience a double hop, or when the signal goes from the ground to the satellite and back to Earth and then repeats the trip to a different satellite before any information is communicated to the receiver. The total delay is about 600 milliseconds, but with a double hop, the latency effect is felt on both ends. While transmissions delays are unavoidable, there are some solutions that can minimize a signal’s travel time so that communication appears to occur instantaneously.
Hybrid Delivery Systems
Today, many corporations have created a hybrid satellite and terrestrial network, says Hossein Sharifi, an expert in satellite communication technology and president of International Communications Group, a satellite consulting firm based in Bethesda, Md., that also offers program management and communication systems technology services. This hybrid delivery system has been adopted by some in the newsgathering field. Stations may use a single hop to move information from the reporters’ location in the Middle East to London, for example, then the data can be carried via fiber optics to New York. “You use only one satellite hop, then the rest travels terrestrially, which would add another 100 milliseconds to the trip,” says Sharifi, also former executive vice president for Hughes Space and Communications International, which is now part of Boeing Co. “It takes 350 milliseconds, which is OK. A lot of companies are doing it this way, which is the best solution for now.”
Another option for moving data from remote locations in Africa, Europe or the Middle East is the Thuraya Mobile Satellite System, which can transfer voice and data using telephone handsets and fixed stations. When using this device, the signal is uplinked to the satellite, down to a gateway and then transmitted across a public network. The Thuraya system suffers from latency issues just as other communication system, but the key advantage is that it can be used for both satellite and terrestrial communications. Just like a standard cell phone service, you would need to purchase a handset priced at approximately $700 and pay a monthly service fee.
Aim for One Hop
Companies have even more options when transmitting data. The first requirement is to find out what type of satellite configuration your system supports, such as mesh traffic or a star network, Sharifi says. Most corporations use a star network where all terminals communicate with the hub and each other through one satellite. While one satellite hop occurs when the terminals communicate with the hub, two hops are needed when they communicate with each other. Other systems are designed for mesh traffic, where all terminals connect to each other with one hop via one satellite. However, he says ground satellite stations called gateways are also used to redirect traffic.
Unlike a star network, mesh networks are ideal when the traffic pattern or signals are uniformly distributed between cities. But in reality, Sharifi says traffic distribution is not uniform in all the different regions covered by the satellite. In the United States, for example, there is much more traffic in major cities than in small cities. Because of the restrictions of mesh distribution, gateways must be used to reroute traffic from low capacity regions to high capacity regions. “As an example, satellite gateways can be installed in places like Nevada, which has no or minimal traffic or rain loss,” says Sharifi. “You can bring the traffic down to Nevada, then route the signal to high density areas like New York through terrestrial lines.”
Using additional satellite gateways balances the system’s capacity, overcoming the issue of satellite capacity distribution. But there is a downside. It will increase the cost while possibly decreasing the quality of voice due to the additional ground equipment that is needed, says Sharifi. While some companies have no choice due to the design of their systems only a single satellite hop should be used. If concerned about satellite latency, two hops should always be your last choice. “You can feel the difference,” says Sharifi, referring to the time difference between a single and double hop. “But sometimes, it’s not about what’s acceptable, it’s about what’s available.”
At other times, satellite latency problems occur with applications that are highly interactive, such as client server software that runs on a laptop and is connected to a server at a company’s main office. These two applications are exchanging data often, which is referred to as “ping-pong protocol, which is very sensitive to the latency involved in a satellite link,” says Eric Siegel, senior analyst at Burton Group, an industry analyst firm based in Salt Lake City.
Consider every time you want to open a remote file. You may only need to click once, yet, before you receive that file, many things need to happen, says Siegel. Maybe your security level or access to that file has to be checked or the system has to verify that the file exists. Although each of these tasks may only take a fraction of a second, they all add up to frustrating delays, he says.
This also can occur when working with custom applications designed by your company or financial programs involving complicated equations or analysis, Siegel says. Although the program resides on your laptop, it has to access your company’s database on the other side of the satellite link. You may need to retrieve information from that database 30 times to complete a single task — the application retrieves some data, realizes it needs more information and then asks for more data. The application also can compare those figures, requests additional information and then make more comparisons. This process continues for a while until the application completes its calculations. Each and every time it goes back to the database for another retrieval, another fraction of a second has lapsed. If the database is located a half-second away, the delay will be noticeable, he says.
Instead of making multiple, simple requests or queries, a company’s information technology department can write a complex query for the database, possibly using procedures stored in the database instead of in the laptop’s software. “It can send the database step-by-step instructions, commanding it to do a computation by retrieving specific bits of data and comparing it to other data,” says Siegel, who says the amount of time it takes a database to retrieve and calculate information is much less than satellite latency time. “So an effective database does the complex operations and gives you the results.”
If the application cannot be redesigned, consider using an application remoting solution like Citrix Presentation Server or Microsoft Windows Server Terminal Services, which can move the entire application into the server room, says Siegel. It is as if a TV camera is in the server room, focusing on the screen, he says. Only computer keystrokes, mouse movements and screen images are transmitted via satellite, so each time an application retrieves information from the database in the same server room, there is no distance involved and data transfer happens quickly. This is an effective solution for any application that is tightly coupled or “ping-pongy” and cannot be rewritten because it is a custom program or may be shrink-wrapped, he says.
Likewise, consider an accelerator box, which speeds up the transfer of large data files as well as the process of opening and handling remote files. If you have default flow controls for bandwidth, sending massive documents will be really slow, says Siegel. More than likely, you will experience a burst of transmission, total silence for a half-second, then another burst of transmission followed by silence. This cycle will repeat until the entire file is transferred.Part of the reason is the default flow control settings for Windows, which will only allow 17,520 bytes of data to be transferred per request, a relatively small amount of data, he says. The system then waits for an acknowledgement that the data has been received before sending more information. That works well with terrestrial lines where there is minimal latency, but satellite latency can make this file transfer “horrendously long” without an accelerator box, he says.
Users also can also try adjusting flow control parameters such as increasing the size of the chunks or blocks of data that can be sent before acknowledgement is required from the other end of the transmission. “People spend a lot of money for a fat pipe or high-bandwidth satellite link,” says Siegel. “Then they discover they can only transfer 35 kilobytes per second per data stream so it doesn’t matter how fat that pipe or link really is.”
However, if the line has errors or the applications make extensive use of remote Windows files, an accelerator box would also be effective. Despite a price tag of more than $3,000, Siegel says it will pay for itself, because it tunes noisy lines, enabling users to fill a pipe or be much more efficient with bandwidth. An accelerator box also can improve the performance of the Windows’ protocol used for accessing remote files.
Even though satellite still has the same latency problems as it did 30 years ago, it has become such a common delivery system that people forget they are using it, says Siegel. “Back then, communicating over satellite was a big deal,” he says. “Not anymore. Yet, people still complain when the performance isn’t that great instead of doing what engineers have been doing for years — going out and buying an accelerator box or other equipment to fix the problem. People simply need to be reminded that when they go over a satellite link, latency problems must be considered.”