In the past 10 years, numerous solutions have been developed to deal with limitations in the leading relational databases. But knowing which offering works best per use case can be difficult to decipher.
Avi’s usual recommendation for Hadoop-type jobs is the MapR distribution. He have found that MapR is similarly priced but offers higher performance and a native NFS interface to Hadoop that can perform at hundreds of gigabits at scale, utilize 24 to 90 disks (depending on CPU and RAM), and allow basic Unix tools to be used for analyzing subsets of data in a more ad-hoc fashion. At the end of his session, Avi will make recommendations based on common use cases, data back ends, and application requirements.