This is a test using large packets, including all overhead (headers and interframe gaps), measuring 20 simultaneous HTTP sessions in each direction; in the case of Island, over four one Gig ports. Transferring large packets takes only about a tenth of the processing power that it takes to transfer small packets, thus we believe the max packet rate using small packets (follow-on test) is a more meaningful test.

This is a test using smaller packets and is a better measure of a router’s overall performance. As a meaningful percentage, usually a third or more of all packets are smaller. Even when streaming large packets, constant acknowledgements back to the other end uses smaller packets. The majority of upstream packets are tiny.

This is a practical measure of maximum connections per second via TCP-how fast you can open connections to websites, and the hundreds of transactions it can take to open a page, especially one with data plus images and ads and animations and updates that are all separate transactions.

This is a commonly used test in the industry that makes it pertinent for that reason; you can see comparative results, given the hardware is the same. A score of 950 Mbps is as good as you can get with 1-gig ports, indicating a saturated Ethernet. This is a test of how much bandwidth you can get through the router downloading a large file from the web. This test counts data only, so the difference between this number and 1 Gig is all the overhead.

All testing of Island and its performance indicators have been conducted internally via PerfTech, Inc. To date, Island has not been tested via third party.

For testing the network performance of Island and competitive products, we acquired Keysite Technologies’ Perfect Storm hardware in our labs and we’re using their BreakingPoint application, a highly reputable testing solution. All performance statistics we report are generated by this solution.