A real-time large-volume content delivery system using globally distributed storage

In the digital content market, with globalization of supply chains and increasing demand for swift provision of content, the need for innovative platforms—capable of transmitting and distributing digital content much more efficiently than the conventional technologies in place on the existing Internet—is increasing. We consider an approach based on a many-to-one content distribution scheme—characterized by distributed storage of largevolume data on multiple storage devices and on-demand gathering of necessary data for browsing—to be highly promising to meet such demand because the dividing of content data leads to the following: smaller bandwidth requirements in the regions near transmission sites, lower probability of excessive loading on storage devices due to concentrated access to a particular content, higher safety against data leakage, and better fault-tolerance. However, due to the increased number of flows characteristic of many-to-one type delivery systems, TCP (especially for retransmission processing) or other conventional traffic control methods become inappropriate for guaranteeing communication quality as these tend to be more complex and to increase control load. In this research, we propose an approach to establish a platform capable of highly efficient transmission quality control taking advantage of the many-to-one type delivery. It employs data redundant technology (by means of error correction code) and virtual network technology, assuming the use of a transmission protocol that does not require retransmission control. The proposal consists of three element technologies (see Fig. 1): “Multi-slice based traffic control technology” that achieves efficient traffic diversion through combined use of multiple routing planes, or virtual networks (i.e. slices); “Redundant encoding and transmission technology for