See http://tinyurl.com/cp57sds for some of the vendors who provide solutions for Big Data Monitoring.
With the traditional monitoring solutions, vendors like BMC (with proactivenet), Netuitive, VMware (with acquisition of Integrien) were focused on identifying brownouts in advance.
Has any vendor in Big Data focused on this? This is still important. What happens if the current node is not performing optimally? Should we not identify non-optimal performance and switch computing to better performing nodes & fine tune the slow performing nodes?