IBM & Anaconda: Reducing Complexity and Risk for Cognitive Development

By Charles King, Pund-IT, Inc.  April 19, 2017

Solutions designed to reduce or eliminate complexity and risk are mainstays in the IT industry, especially when it comes to emerging technologies. The short and long term intentions of such efforts are pretty straightforward; by doing so, vendors aim to ease customers into trying on new offerings in hopes that the experience will result in future sales opportunities.

That’s been the case with a number of efforts IBM has initiated around its Power Systems and technologies, including open sourcing its POWER processor architecture and helping to found the OpenPOWER Foundation with vendors interested in leveraging Power across a range of innovative new enterprise and data center solutions.

But it’s also apparent in the company’s recently announced partnership with Continuum Analytics to offer its Anaconda Open Data Science platform on IBM’s POWER-based Cognitive Systems and to integrate Anaconda with the PowerAI distribution for machine learning.

Evolving Power

Why all the focus on POWER technologies? From a technical perspective, the robust memory, throughput and other capabilities that make IBM Power Systems mainstays in thousands of enterprise customers’ datacenter (and enabled the company to capture and maintain a massive competitive advantage in high-end server sales) are highly complementary to machine and deep learning processes.

In addition, IBM is continually proactively evolving POWER-based silicon and systems on its own and with the assistance of strategic partners. For example, last year’s introduction of PowerAI leveraged deep integration with NVIDIA’s NVLink interface. That enables a CPU-to-GPU interface with NVIDIA’s Tesla Pascal P100 accelerators that significantly boosts machine learning, deep learning and advanced analytics applications.

Some efforts are more broadly strategic. For example, in February IBM formed a new Cognitive Systems BU headed by SVP Bob Picciano who previously led the IBM Analytics organization. Another recently announced effort, the OpenPOWER Machine Learning Work Group (OPMLWG) finds Canonical, Cineca, Google, Mellanox and other vendors collaborating to define frameworks for machine learning solutions using the IBM POWER architecture and OpenPOWER ecosystem technologies.

Anaconda: Squeezing a “goldmine” out of basic data

Where does Continuum Analytics’ Anaconda fit into all this? The Anaconda platform is designed to simplify package management and deployment processes in numerous large-scale data processing, predictive analytics and scientific computing scenarios. According to Travis Oliphant, the company’s co-founder and chief data scientist, Anaconda tools “Can identify patterns in data, uncover key insights and transform basic data into a goldmine of intelligence.”

Those capabilities have resulted in widespread interest in Anaconda among developers and data scientists—with over 16 million downloads of the platform to date. Offering Anaconda on IBM’s POWER-based Cognitive Systems and integrating it with PowerAI will also enable developers to access and explore PowerAI Platform libraries, and deploy them on Anaconda Enterprise.

Final analysis

Often times, interest in evolution focuses on “ah ha” moments and giant leaps ahead. However, the practical effects of evolutionary advancement are typically subtler with small yet significant, sustainable steps forward being the norm.

That sort of progress is clear when one studies IBM’s efforts related to POWER technologies. The company continues to advance on its own, for example in development of its upcoming POWER9 processors and resulting solutions. But it is also making broad progress with the enthusiastic assistance of numerous partners, including the 300+ current members of the OpenPOWER Foundation.

This new partnership is a more pointed effort aimed to mutually benefit both IBM and Continuum Analytics. However, doing so should also substantially enhance the two companies’ respective Cognitive Systems and Anaconda Open Data Science platform.

Most importantly, the effort should also help inspire interest in machine learning, deep learning and cognitive solution development among businesses and other organizations. In many cases, we open doors today to allow or encourage opportunities tomorrow. That is clearly one of the goals IBM and Continuum hope to achieve with Anaconda.

© 2017 Pund-IT, Inc. All rights reserved.