Two luxury cruise lines faced the same satellite-delay challenges for shipboard Internet access and came up with different ways to address the unavoidable latency on those ship-to-shore connections and keep passengers happy.

In one case, Regent Seven Seas opted for deploying Riverbed Steelhead WAN accelerators on each ship, and another at the satellite downlink in New Jersey, to attain higher throughput over the 256kbit/s burstable satellite circuit.

In the other, Crystal Cruises installed a single F5 Big-IP device at the satellite down station in California, which provides a 512kbit/s connection to Crystal ships.

Both lines say the WAN gear has boosted performance significantly enough to cut the need for buying more bandwidth from the satellite provider, MTN Global Networks, a division of SeaMobile.

In the case of Regent Seven Seas, the 256kbit/s bandwidth can spike to 512kbit/s when needed. All passenger and crew have Internet access, email, cell phone service, and data to and from the ship goes over the satellite. Because of the vast distances involved between the satellites, ships and ground stations and the speed limit on the radio frequencies, the connections experience latency of more than 600 milliseconds.

Rather than use Riverbed gear directly on passenger traffic, Regent Seven Seas uses it to reduce the amount of bandwidth it takes to transmit administrative traffic. Through an arrangement with its satellite provider, passenger Internet bandwidth management is outsourced and already includes WAN optimisation techniques, says Vincent Cirel, CIO of Regent Seven Seas.

By optimising the administrative traffic with Riverbed gear, more of the satellite bandwidth remains available to passengers for Internet access, he says.

“The property management systems, the financial systems, just the overall maintenance activity that has to take place for remoting into servers onboard the ship - there are very, very high bandwidth requirements particularly for doing file transfers,” Cirel says. “Without the deployment of the Riverbed devices on the admin side of the network we would be crowding out the passenger usage requirements, and it would mean we'd have to monitor things a lot more closely than we do.”

For instance, if a ship were at sea with no landing scheduled that day, up to 300 passengers might be logged on to the Internet at the same time. “We would have to look at the itinerary and decide if we prioritise the admin traffic, what is that going to do to the passenger traffic?” he says.

With Riverbed optimisation, that kind of time-consuming decision comes up less often. “By deploying this type of technology, we remove a lot of that manual analysis,” he says.