Researchers at the Massachusetts Institute of Technology are building a prototype interactive database for big data applications that could deliver information from huge datasets almost immediately. Currently, data that needs quick analysis is stored in a computer’s main memory, or dynamic random access memory. However, the datasets of today are too large for that.
The researchers are developing a system that uses multiple nodes across an Ethernet network. The Ethernet standard has grown to encompass new technologies as computer networking has matured; a key element of Ethernet’s success is its ability to coexist with existing network installations.
Charles W. and Jennifer C. Johnson Professor of Electrical Engineering and Computer Science, have developed BlueDBM or Blue Database Machine, based on a network of flash storage devices. In BlueDBM each flash device is connected to a field-programmable gate array (FPGA) chip to create a node. The FPGAs not only control the flash device, but also perform processing operations on the data itself, Jun says.
FPGA chips can be linked together using a high-performance serial network, which has a very low latency, or time delay, meaning information from any of the nodes can be accessed within a few nanoseconds. Using multiple nodes allow the team to get the same bandwidth and performance from their storage network compared to more expensive machines, he adds.