Quick thoughts about yesterday’s HortonWorks / SAP reseller announcement
Yesterday, there was an announcement about a new agreement that represents the next stage in SAP’s BigData strategy regarding Hadoop.
The agreement enables SAP to resell Hortonworks Data Platform (HDP), the industry’s only 100-percent open source data platform powered by Apache Hadoop. It will allow users of SAP(R) software with a simple way to incorporate Hadoop as a complementary component of their data architectures to enable a broad range of new applications as well as enrich existing ones with additional data sources.
I was reading the various press reports about the deal and I discovered a few points that I want to explore in more detail.
HANA just for “traditional sources”?
The description of deal from HortonWorks also includes the architecture behind the cooperation.
What I found curious was the apparent restriction of HANA to “traditional” data sources. The “new” sources are the focus of Hadoop.
This distinction is also expressed in other parts of the blog:
By using SAP HANA and Hadoop together, customers get the power of instant access with SAP HANA and infinite scale with Hadoop. This gives SAP users a broad range of options for storing and analyzing new types of data and the ability to create applications that can uncover new business opportunities from vast amounts of data that would not have been previously possible.
HANA is positioned, however, for these “new” data sources as well. Here are a few examples:
- Geo: Since SAP HANA SP6, the native spatial processing is available in HANA. There is also an agreement with Esri in this area.
- Sentiment: As this blog demonstrates, sentiment analysis is also possible in HANA
- Sensor: The use of sensor data with HANA is being demonstrated by a few start-ups. HANA is also an integral part of SAP’s Internet of Things strategy.
Thus, it is evident that SAP and HANA aren’t really restricted to those “traditional” data sources.
I found SAP’s explanation behind the cooperation more accurate:
“Integration with Apache Hadoop is part of SAP’s overall strategy to provide valuable insights across a continuum of data from the efficient storage of massive amounts of cold data, to petabyte-level storage of warm data to real-time and streaming data analysis,” SAP said in a statement. [SOURCE]
What is more important is not the source of data stored in each platform but rather how the data is stored : “warm” vs “cold”.
SAP support for Hadoop
Another part of the agreement deals with Hadoop support:
SAP has entered reseller agreements with Intel and Hortonworks to resell and support the Intel Distribution Apache Hadoop and the Hortonworks Data Platform with SAP HANA, expanding existing partnerships with both vendors [SOURCE]
This general statement about support was interesting. Did SAP have enough expertise in its support staff to deal with these new areas? I looked and SAP didn’t have any committers on the Apache Hadoop project.
If a new Hadoop-related support staff was being created, I expected a number of Hadoop-related job openings for support positions.
A search on SAP’s job site, however, shows no such increase:
I finally found some support-related details that shed light on the topic:
SAP will provide level 1 and 2 support for customers and will be backed by the Hortonworks support team. The Hortonworks support team, experienced in supporting large and complex Hadoop deployments, is backed by the core engineers and architects of Apache Hadoop enabling SAP customers to get world class global support backed by deep knowledge of this emerging technology. [SOURCE]
I find this cooperation between the two support organizations the best solution in that it allows each organization to concentrate on their particular area of expertise.
I assume that this sort of support structure is similar in other commercial relationships where SAP acts as a reseller.
Note: I wasn’t able to find any more details about support related the reseller deal with Intel.
Thanks Richard for this writeup. I think it will be very interesting to see how SAP will evolve in this new Partner space. For the moment it is Intel & Hortonworks but there is many others which should probably not be neglected since they have their own traction in the market like Clouder & IBM BigInsights.
I am very curious to understand better the support models and the further addeded value of buying an "opensource" solution from SAP vs directly from any of the vendors in the market.
Hi Richard,
Thanks for the writeup. I just wanted to help clarify the interpretation of the simple diagram above.
The key point is that the data systems layer in a modern data architecture needs to be able to handle both traditional and new sources of data AND (to your point) be able to manage that cost effectively over large volumes and long periods of time (ex. hot, warm, cold scenarios). The placement of the arrows are not meant to imply which system handles which type of data; the point is the middle layer needs to handle both.
As you point out, HANA is pretty flexible in its ability to handle a wide range of data types. By enabling HANA and Hadoop to work well together, you get a data architecture that handles instant access and infinite scale quite nicely.
Also, you cover the topic of support. The point here is to enable customers to work with their trusted vendor (SAP) even for Hadoop support. Our relationship with SAP will enable a transparent handoff for Hadoop specific issues that need to escalate to the Hortonworks experts who are uniquely qualified to handle them. Given the fact that we have over 100 engineers who do all of their work in the Apache foundation projects (with 21 of them being active committers to the Apache Hadoop project alone), we feel SAP customers will get the world-class enterprise support they demand in the manner they are most familiar with.
I hope that helps clarify.
Shaun Connolly
VP Strategy, Hortonworks
Shaun,
Thanks for responding to my blog.
I concur that the middle layer depicted in the diagram is the most important. I was just irritated by the message that readers might receive when viewing the chart. As the famous saying goes - "A picture is worth a 1000 words".
Regarding the issue of support, I also agree that HortonWorks engineers are probably better suited to deal with Hadoop-related issues.
D.
Thanks for this information Richard!
Regards,
vivek