Hadoop offers large enterprises a way out of their Big Data processing challenges but the Java-based programming framework but comes with a few pain points as well.
Not the least of these is Hadoop’s single point of failure – when one NameNode in a cluster goes down it takes the entire cluster with it.
Restoring crashed clusters take time, which makes many C-level executives skittish about considering Hadoop for mission critical applications.
David Richards, CEO and chairman of California-based software company WANdisco Inc.says the firm’s peer-to-peer, data replication technology which keeps data and files synched throughout data centres can be tweaked to make Hadoop “bulletproof” for the enterprise, according to an article in Siliconangle.com.
He said WANdisco and its recently acquired Hadoop developer firm AltoStor will soon release a solution called the Non-Stop NameNode which will guarantee that Hadoop will stay up all the time.
With WANdisco’s peer-to-peer strategy, each node in a cluster can service read and write. Administrators can take down a node for maintenance and a peer node will pick up from where it left off. If it works, this will have huge implications for mission critical applications.
Read the whole story here