Hadoop Redux – Trends and New Milestones

Share

It’s a whirlwind week in the Hadoop ecosystem with Strata/Hadoop World, new research from Sand Hill Group and key milestones such as YARN to support execution engine alternatives to MapReduce.  Let’s break it down briefly.

First, at Strata, we are demonstrating our joint Reference Architectures with Hortonworks, leveraging our ParAccell Dataflow for Hadoop.  One targets graphical development for very high-performance ETL/data quality and the other  is geared for graphics development of advanced (non-SQL) analytics on Hadoop.  Both are certified for HDP 2 as well as YARN.  Actian CTO Mike Hoskins and Hortonworks CTO Ari Zilka also participate in a joint panel hosted by Ovum analyst Tony Baer on using YARN-ready technology to manage and analyze data on the HDFS cluster in a visual framework – without needing MapReduce or Java programming.  Actian’s view is that YARN is a major milestone for the Hadoop community.

The real goal of YARN and our other Hadoop initiatives is to deliver value more quickly for the business from data in Hadoop.  Check out research from Sand Hill Group released today, titled Do you Hadoop? - A Survey of Big Data Practitioners shows that Hadoop has, as MR Rangaswami’s summary neatly puts it, “moves well beyond the ‘science project’ status.”  As we noted in a related announcement last week, the excitement around successful adoption of Hadoop for enterprise-scale production initiatives is accompanied by some genuine friction points – data extraction challenges, design-time hurdles and poor run-time performance.  And there are other challenges surfaced in Sand Hill Group’s research: for 46.7% of respondents, the top challenge is knowledge and experience with the Hadoop platform, while 20.7% percent cite availability of Hadoop and Big Data skills and 6.7% percent struggle with the amount of technology development and engineering required to implement a Hadoop-based solution.  Actian’s focus on delivering a graphical interface to elegant data preparation and analytics solutions that run natively on Hadoop – with amazing design-time and run-time performance – aims to tackle these challenges head-on.  You no longer need specialized MapReduce skills to leverage the wealth of Hadoop-based data.  Further, the ParAccel Dataflow for Hadoop engine brings pipeline parallelism in our native Hadoop execution under YARN, up to 500x faster than MapReduce.

As Actian CMO Ashish Gupta put it recently, we are in a time of tectonic shifts in the data analytics market.  As he notes, “With our platform, you can design an end-to-end process, from ETL through data prep to highly sophisticated analytics, with a drag-and-drop GUI to design the flow, including prebuilt data preparation and analytic functions you can stitch together. And you can write your own specialized functions to meet your unique business needs.  When you want to dig in with deep analytics, our powerful ParAccel SMP and MPP databases can scream through massive volumes at warp speed.”

There’s been some justifiable concern lately that Big Data is over-hyped.  With exciting inflection points in Hadoop adoption, significant community milestones such as YARN, and burgeoning business value-focused offerings from Hadoop ecosystem players, I feel good about our ability to deliver hard-hitting, enterprise-ready big data analytics solutions for customers worldwide – with true ROI through game-changing competitive advantage, fundamentally better risk management and entirely new business service models.

Join the conversation! Tweet us at @ActianCorp #AgeofData

About Alison Raffalovich

Alison heads Actian Corporate Marketing and Communications and loves to talk about the cool innovations from Actian partners and customers in the Age of Data.

View all posts by Alison Raffalovich →

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>