Today, the LOFAR central processing team proudly announces that the LOFAR phase II cluster is operational. Over the next couple of days they will show the first observations stored and reduced on new hardware recently installed in Groningen. They have added some 100 hybrid compute and storage nodes, containing 2448 cores, 6.4 terabyte of main memory and just over 2 petabyte of disk storage to the already impressive list of hardware. The entire cluster is in theory capable of performing just over 20.5x1012 floating point operations every second, four times that of the previous cluster.
This new cluster is designed to handle far greater bandwidths than the previous cluster. With this hardware LOFAR should (eventually) be able to handle up to 80 Gbps of data streaming from the Blue Gene/P supercomputer, allowing LOFAR to observe with more beams, at higher time resolution and with more frequency bands than previously possible. According to one enthusiastic pilot user LOFAR can now do with one node what until now required the entire cluster.
The picture above shows a part of the cluster. What can't be shown in pictures is the tremendous amount of work that was done, by personnel from both the University of Groningen and ASTRON, to get the operating system and software working in record time.