This Does Not Compute: Oops, Cray + Hadoop

Sometimes “innovation” simply does not compute.  Cray’s supercomputers would seem the last word in secure computational power. Apache Hadoop, on the other hand, has, among other inherent vulnerabilities, design errors that make it vulnerable to information disclosure, according to Security Focus and other sources (identified by Cloudera). In fact, as quoted by Data Informed, James Vogt of Zettaset has pointed out that in distributing big data jobs across clustered computers, Hadoop “shares data and conducts batch processing of data across nodes,… [creating] a different set of security challenges that you can’t really address with perimeter security.” According to Information Week, Cray promotes Hadoop distribution as bringing “greater security” to Cray’s clusters of half a million dollar, and up, supercomputers. Somehow it seems that for healthcare, finance, and other highly sensitive sectors the combination actually puts designed-in vulnerabilities on steroids and kicks security holes all through Cray’s supercomputing solution. A Cray cluster coupled with a suite of open source software applications written by a community not committed to high security performance would seem destined to lead to unintended consequences —and most unfortunate lessons to be learned. Again. Has hype put the “Cloud” right on the edge of super-disaster, yet again?

© 2013 Copyright. S Caroline Schroder. All rights reserved.