In CDH3 we were having these NameNode, SecondaryNameNode only, that time we faced SPF(single point of failure),If NameNode got down total cluster will be down. But Now in CDH4, this problem is eliminated by using alternate NmaeNode (standby NameNode).
In CDH4 people they have introduced two NameNodes, and the only one will be in service and the another one will do the work what secondarynamenode does, and comes into service when the working namenode goes down. This feature is called NameNode HighAvailability (NNHA).
answered By  0
Name node and jobtracker are the single points of failures but we have got an alternative from CDH4 and YARN(mapreduce2) so if u wil use CDH4 and YARN u r Namenode and Jobtracker are wont be singlepoint of failures