- End-user does not need to configure performance.
- Can be deployed to any type or speed of network.
- Instantly adjusts to changing conditions.
- High performance across the full spectrum of network bandwidth.
- Does not flood the network: automatically shares with others.
- High performance with round-trip latency up to 20,000 milliseconds.
- Tolerates congestion loss rates up to 50%.
- Data keeps flowing during packet loss recovery.
- Adjusts flow as traffic conditions change.
- Simple, modular interfaces.
- Fire and forget data transfer.
- True source-code integration.
- No external dependencies (e.g. no Java, no Cygwin, no DLLs)
- Combined libraries total less than 1 megabyte.
- Can work beside and with legacy transports.
- Uses just one UDP port: no TCP.
* "Real-world" means real customers using our software on their production networks and comparing the results to their previous data transfer methods. It does not mean lab-tests, emulators, or comparisons to ancient, crippled TCP stacks. Anybody can "dial-a-result" to claim 100x or even 1000x with canned tests. We don't.