Signiant said it has developed a new data transport system for files and live media streams based in part on machine-learning techniques for analyzing network history and reacting to congestion issues.
The file-transfer specialist has applied for a patent on what it calls a new “intelligent transport architecture” that replaces Internet-standard TCP transport, which has significant overhead related to verification of data packet delivery. Instead, the system uses either parallelized delivery through multiple TCP connections or a proprietary, accelerated protocol based on the more efficient UDP, depending on network conditions.
A machine-learning algorithm configures those operating parameters based on past history, which Signiant says eliminates the need for manual tuning and also has the potential to improve over time as more data is collected for analysis. Signiant’s UDP acceleration is enabled by analyzing latency and packet loss, including tracking the rate of change of both, in an effort to distinguish between edge and core network congestion.
Signiant said the new transport system can reach transfer speeds in the Gb/sec range for transfers between sites or to and from cloud servers. It has already been deployed in Signiant’s cloud-oriented Flight file-transfer offering as well as its recently announced Jet system for automated file transfer. The company claims its acceleration technology can transfer data at up to 200x faster than standard TCP, especially in high-latency situations.
“As networks evolve and increase in bandwidth, many incorrectly assume that acceleration technology is no longer required,” said Signiant CTO Ian Hamilton in a prepared statement. “Network bandwidth and network throughput are not the same, and Signiant is able to provide even more value on higher bandwidth connections. Our new intelligent architecture supports all types of transfers, including site-to-site, site-to-cloud and cloud-to-cloud transfers of both files and streams with no constraints on file size or type.”
Signiant will be exhibiting at booth SL 10216 at NAB Show 2019.