JERSEY CITY, N.J. July 26, 2011
However, the conventional techniques for adjustable bandwidth network physical layer connectivity fail in this critical aspect: conventional systems for adjusting the bandwidth rates operate at orders of magnitude slower pace than what is required in order to deliver the business value of rightsized network connections. To see why, it needs to be realized that network data bursts arrive in sizes of individual data packets, ranging typically from 64 bytes through thousands of bytes, and lasting for as short time as less than a microsecond. To keep the physical layer connection capacities optimized according to these packet data bursts, the control loop for adjusting the connection bandwidths needs to operate at the same time and bandwidth granularity as is the case with these data bursts to be accommodated. The conventional techniques however operate at completely different granularity, e.g. allocating full 1/10Gbps links for periods measured unavoidably in several seconds due to the delays of traditional, server software based, control systems. As such, though the bandwidth should be adjusted at microsecond timescale in minimum packet size units such as 100 bytes or less, conventional systems are limited to allocating bandwidth in amounts measured in several Gbits — at millions of times too coarse granularity.
The US patent #7,986,713 delivers the critical innovation of automatically optimizing network physical layer bandwidth allocation at microsecond timescale in units of less than 100 bytes. The innovation involves implementing network control plane in hardware logic together with the data plane, to keep the control loop in synchrony with packet by packet data plane dynamics. The result is an Adaptive-Mesh network architecture providing continuously rightsized, transparent, secure and high performance physical layer connectivity between the packet switching nodes interconnected.
SOURCE Optimum Communications Services, Inc.