Hortonworks and the Age of Data

By Charles King, Pund-IT, Inc.  March 9, 2016

Evolution and growth are common in organizations of every sort but especially those in the IT industry. That’s partly due to the nature of technology itself, which tends to get cheaper and easier to produce over time, but competitive issues also come into play. Once a company can make a living, it attracts others who want a piece of that same pie. The process is often like dealing with ants at a picnic, though at other times it’s more akin to a visit from a hungry biker gang. In any case, sitting still and exuding a Zen-like calm isn’t an option unless you prefer a life of poverty.

But how do you get to that new place? The path forward is most often linear, with vendors expanding their well-known strengths into new areas and capabilities. Think of VMware leveraging its hypervisor skills from workstations to servers to other associated data center technologies. But it can also involve leaps forward for both the vendor and the industry, like enterprise storage leader EMC’s acquisition of VMware or PC-pioneer Dell’s decision to become an end-to-end systems vendor.

In every instance, it is critical for a vendor to believably communicate the reasons behind its proactive evolution, the assets, skills and strengths it brings to the journey and the likelihood of arriving at its final destination. Otherwise it risks alienating customers, partners and allies and inspiring them to walk away.

Last week Hortonworks hosted its first annual analyst day in San Francisco that addressed these and other issues. Let’s consider what the company had to say and how well it succeeded in describing its plans and ongoing evolution.

Hortonworks’ Who’s Who

If you’re unfamiliar with Hortonworks, the company was founded by a group of Yahoo! and SpringSource engineers deeply involved in Apache Hadoop, the open source software framework that powers thousands of big data projects and strategies. The company was founded in 2011, went public in 2014 and now boasts over 800 customers, including major players in financial services, telecom, retail and automotive, and 1600+ partners.

Apache Hadoop has been at the epicenter of big data developments roiling the IT industry for a decade. In fact, the 10th anniversary of its launch was early last month. Leveraging the Hadoop Distributed File System (HDFS) and the MapReduce processing system across clusters of commodity servers results in a hugely scalable solution for crunching massive volumes of data. But Apache Hadoop and its supporting community haven’t been sitting still.

Hadoop has helped inspire and evolve over 20 related Apache projects, including Hive (data warehousing), HBase (non-relational database), Spark (a high performance cluster framework), Kafka (message broker), Storm (a distributed computing framework written in Clojure), Ambari (simplified cluster management tools) and NiFi (integration and dataflow automation). As a result, Hadoop has become a shorthand of sorts for the modern data architectures that are changing the ways people think about and work with information.

How Hortonworks has successfully contributed to and leveraged these developments were detailed by a Who’s Who of Hortonworks executives, including CEO Rob Bearden, CMO Ingrid Burton, CTO Scott Gnau, founder Arun Murthy and VP of strategy Shaun Connolly. The narrative flowed naturally, positioning the company’s individual and collaborative efforts in the natural evolution from what Connolly called “the age of RDBMS (led by Oracle) to the age of the Web (and rise of open source) to the age of data where open source has become the norm.”

From big data to all data

The massive volume and variety of information along with its mind-boggling velocity of growth (doubling every two years) have led to two areas where Hortonworks intend to lead. The first is to provide the tools and services organizations need to effectively manage complex information assets. The second is to expand beyond being “a bellwether of big data” to supporting “all data” types, use cases, applications and solutions.

That point was central to one of the major announcements it shared—the general availability of the Hortonworks Data Platform (HDP) 2.4 which has been designed both to stabilize the platform’s core capabilities and to provide easier access to innovative “extended” features and Apache projects. This may seem somewhat contradictory, but it reflects the degree to which Hortonworks’ solutions have been adopted among both mainstream enterprises and technologically forward-looking businesses.

As a result, the company plans to make future HDP updates available in two forms: 1) core HDP services, and 2) extended HDP services. This may seem “much ado about nothing” to some readers, but by altering the cadence of its product updates, Hortonworks will allow customers to proceed and progress at their own rates, a critical point in a highly flexible and rapidly evolving platform like Hadoop.

By doing so, the company has taken a fundamental step in the journey from being a developer of highly innovative technologies to become a vendor that places the needs and preferences of customers first. That same point was driven home in other announcements, such as Hortonworks’ vision for Connected Data Platforms which includes new management capabilities for data both in motion and at rest. It was also highlighted in new functionalities for Apache Spark 1.6 and Apache Ambari 2.2.

Customer and executive discussions

The depth of Hortonworks’ efforts was also apparent in 1:1 meetings I had with company partners and executives. For example, Mike Peterson, VP of cloud and data platforms at Neustar, a telecom information and analytics business, noted the significant benefits of its Hortonworks relationship. In 2004, the company was a “100% Oracle shop” running on Sun Microsystems hardware. While Neustar still uses Oracle for transactions, most of its analytics services leverage HDP running on commodity servers and storage.

To support its information and data registry services, Neustar once captured less than 10% of its network data and retained it for just 60 days. With HDP, it captures 100% and retains it for two years. That considerably deepened the analytics insights Neustar provides customers, but working with Hortonworks’ open source-based portfolio and commodity hardware also provided the company considerable savings on hardware upgrades and software license fees.

As Peterson noted, Hortonworks helped Nuestar achieve its goal of delivering “analytics over multiple geographies,” but he’s also impressed by its directional sense: “They and the Apache Project seem to know where they need to go.”

Another interesting viewpoint was offered by Hortonworks’ VP of global channels and alliances, Chris Sullivan who’s been with the company for just over five months. Sullivan’s professional resume is impressive. Prior to joining Hortonworks, he held similar positions at VCE (where he was the 30th employee hired) and Cisco. His experience at VCE seems especially germane since, early on, the company’s converged infrastructure solutions and services weren’t well understood by many customers and partners.

Hortonworks’ aim of leveraging its deep technical skills into a larger market position is, “very comfortable for me,” Sullivan joked. But he was deadly serious about the need to create value with partners and to execute effectively at a local level. He especially likes the “purity of the Hortonworks’ business model” and noted that, “the use cases, validation and vertical applications are all there.” Most important is the openness of partners and customers to the company’s viewpoint. “Is the market ready for Hortonworks’ message? Emphatically, yes.”

Final analysis

So how were Hortonworks’ presentations and strategy viewed by analysts at the San Francisco event. Positively, overall. Company representatives were thoughtful and well spoken, and a few (like Shaun Connolly’s presentation on the continuing value of open source) seemed inspired. Hortonworks’ narrative, like its current journey, is essentially linear, requiring few, if any, leaps of faith. In other words, no suspension of disbelief was asked for or required.

That said, a few rough edges need attention, like the company’s product naming schema that relies on Apache’s project/version numbering. That’s fine for the data center staff and developers who actually make the stuff work, but it’s likely to inspire little but confusion and heavy eyelids among non-technical executives and managers. If Hortonworks hopes to expand its marketplace position, it needs to communicate effectively with those audiences, too.

Beyond that, the San Francisco event cast a helpful and hopeful light on a company with a significant roster of achievements that appears poised on the edge of greater success. As enterprises and the greater market inevitably move from the big data present to an all data future, Hortonworks is likely to be a constant companion and trustworthy guide.

© 2016 Pund-IT, Inc. All rights reserved.