IBM hooks up with Hortonworks to Hadoop Big Blue products
But customers welcome to taste IBM's own-brand Hadoop sauce...
+Comment Cognitive Watsonian and mainframer IBM is making Hortonworks available for its Elastic Storage Server (ESS) and Spectrum Scale products.
ESS is an integrated bundle of POWER8 servers, storage enclosures, disks and Spectrum Scale software. The GPFS parallel file system software was renamed Elastic Storage and then rebranded again as Spectrum Scale. Both ESS and Spectrum Scale are called software-defined storage products* by IBM.
IBM now has an agreement with Hortonworks to certify the Hortonworks Data Platform (HDP) with Spectrum Scale on Power and x86 ESS server hardware. HDP, which is a Hadoop distribution using a centralised architecture, was recently made available on POWER8 server hardware, and this announcement moves things on a step.
This will be the first IBM storage offering certified for Hortonworks' Hadoop, according to Big Blue, which makes us wonder why IBM has been so slow (but see the note below about IBM's own Hadoop distribution).
The deal suits both parties. Hortonworks needs server and software support for its Hadoop distribution and IBM can use HDP to help sell its ESS products into Big Data analytics prospects wanting Hadoop.
Ed Walsh, general manager for IBM Storage and Software Defined Infrastructure, has a canned quote and he or his PR flack writer couldn't resist putting a cognitive spin on this: "Every organization is becoming a digital organization. With this announcement IBM is delivering a powerful platform to extend the use of data and for cognitive applications."
That provides a high marketing buzzword score as Walsh got the "digital organisation" term in there too.
Buzzword sniping apart, IBM customers can analyse enterprise storage data directly instead of having to copy it to a separate analytics platform first, which is good news. IBM also reminds us HDFS makes three copies of data whereas ESS uses the more space-efficient erasure-coding for data protection.
It also claims ESS, with parallel file access, can be faster than distributed HDFS deployments, though not when ESS uses a cloud back-end tier – but that, it says, can bring cost benefits.
Lastly, in its announcement, IBM wants us to know that its own Hadoop and Spark distribution and analytics software is also an option for customers using Spectrum Scale Storage.
El Reg could well imagine Hortonworks would be pissed off at reading this in an IBM press release extolling IBM product certification with Hortonworks's HDP product, and justifiably so. ®
*Well, actually software-defined storage solutions but we couldn't bring ourselves to use that horrible marketing-speak term.