Should SAP open source HANA?


On and off back channel discussions about open sourcing SAP HANA come back into view as IBM gets with the open source theme.

SAP HANAThere may well be a fair few raised eyebrows at SAP with this headline but there is a seriousness attached to it that is worth exploring. I am not the only one who has thought of this. A few years ago, when SAP Mentor/developers were getting a good amount of attention from SAP’s technical folk, the licensing topic was never very far from people’s minds.

At one point, the question of whether SAP should open source HANA was very much in the air but not considered seriously by the company. They thought HANA was far too valuable a property to toss into the open source arena. Regardless of where you stand on this topic it was hard to argue against the commercial considerations of the time, especially given the implied promise that HANA made.

SAP’s open source chops

In the interim, SAP has quietly gotten alongside some of the larger open source communities, committed code into various groups and generally shown itself to be a good actor in the open source world. It didn’t go un-noticed that SAP had a booth at the recent Hortonworks Hadoop event. It partners with Cloudera – again Hadoop, has a role with Cloud Foundry and a long term relationship with Eclipse. Then there is Open UI5. So you can readily make the argument that SAP has got a fair degree of open source credibility. So why bring this topic up at all?

As I’ve spent more time in the commercial open source world, it has become clear to me that the majority of business application development is going to happen out of open source components. SAP recognizes that already to a limited degree.

Bring on the hardware

That same thinking is creeping into the hardware space. Check the weekend piece about Facebook pushing its OCP into open source. While Business Insider is all over this as if the world is going to come to an end for the likes of Cisco, this is not as apocalyptic as it might sound. Having said that, I’m thinking about how the server market is shrinking, how prices are under pressure, how commodity hardware is really the only way you can get to web scale at reasonable cost. On the other hand, I’m also thinking about how another Facebook invention, Cassandra, has become a focus for software development. Facebook saves many millions of dollars by working on open source projects of this kind.

IBM is in

Then there was the weekend announcement that IBM is plunging deeper into open source with a commitment to Spark, yet another of the Hadoop related projects. In this case, Spark really is more of a science project than product per se but then it has one of the largest, if not the largest communities inside the Apache network Something must be driving of interest. Here’s your starter for 10:

[IBM]…will embed Spark into its analytics and commerce platforms, and to offer Spark as a service on its Bluemix development platform. IBM also is expected to announce that it will assign more than 3,500 engineers and developers to work on Spark-related projects around the world, and that it will donate its IBM SystemML machine learning technology to the Spark open-source ecosystem. It also is expected to launch a Spark technology center in San Francisco and help educate data scientists and engineers in the use of Spark on a mass basis.

Given IBM’s reach, the commercial considerations are obvious: IBM + new hotness (Spark) = bums on consulting seats and lots of code being written aka kerrching at the IBM cash desk. In short it allows IBM to ride open source coat tails while protecting and enriching its business model.

Yes yes well this is all very nice but where does this SAP HANA thing fit in?

So what about SAP HANA?

HANA is the database all SAP technology will eventually sit upon. That means, if you buy SAP then you buy HANA. If you already own SAP you eventually buy HANA. I think that’s at best a sustaining proposition rather than a growing value proposition unless you are in a position to solve the kinds of problem for which HANA was originally designed. If anything it suggests that all SAP is really doing is seeking to grab maintenance revenue from Oracle, Microsoft and IBM to itself. Those other companies supply the databases upon which SAP technology has traditionally worked.

For its part, SAP could hardly be said to be coming into market with anything fundamentally new, unless you count S/4 HANA in that bucket. And I for one do not. Even so, others think that SAP’s product portfolio is adding new solutions at a fair clip.

All of that means SAP is devoting resource to building database technology and sidecars but not doing that much of what it is best at: solving big hairy problems. At least not as far as I can tell.

Like every other software development organization, SAP has limited resources. If it was to drop HANA into the open source community, I am betting that it would see much wider adoption and become a focus for accelerated development from the point of view of fresh eyes.

That in turn would allow for the acceleration of use case development beyond ERP that itself would provide a rich seam of product SAP could take to itself but which doesn’t exist today. It would also allow the ecosystem to flourish in ways that are not clear today given that SAP is pushing as much as it can onto cloud platforms.

Such speculation doesn’t come without problems. There is history here with MaxDB that SAP at one time jointly developed with MySQL. But that kinda died on the vine. Then there is the question of whether HANA revenue – as in real license sales –  is a significant contributor to top line revenue. That used to be fairly clear but now is very difficult to determine. The common answer of ‘HANA is everywhere’ doesn’t really address the question.

My take

  1. We consistently hear that customers take one look at HANA and baulk at its cost, largely because the business case is perceptually difficult to make. It’s not just the license but the hardware specifications that hobble making standing up a HANA case. Open source solves a good part of that problem. Intel for instance has no trouble working with Hadoop in its chip designs. It has worked with SAP on HANA. How much further could that partnership extend if Intel thought there could be additional revenue from an expanded customer base? How fast would smart hardware designers review HANA specs and come up with something truly innovative like OCP and its spawned products?
  2. Loyal SAP developers who have been muttering about open source for years would likely find a new reason to continue nailing their colors to the SAP mast. Those folk are pretty vocal and world class advocates for a company they believe in. Why not use open source to further exploit that loyalty and get code into the bargain? Even if you think that open sourcing doesn’t change any of the internal development dynamics, it has to be attractive.
  3. The biggest hurdle is the internal perception that HANA represents the company’s crown jewels. I totally get why that would be the case but it isn’t the only answer to many of the issues SAP customers face. And in any event, in going open source, SAP would not lose a penny in consulting or maintenance.
  4. Taking SAP HANA off the price list removes a pressure and distraction for sales. That has to be a net good because it means they can concentrate on the value add applications that SAP has always envisaged as the real prize for customers.

Disclosure: SAP is a premier partner at time of writing

    Comments are closed.

    1. says:

      Brief thoughts, not necessarily limited to SAP, but surrounding OSS more generally. I was an advocate for the Red Hat model due to the market failure at the time, and also support components of OSS particularly in areas that either are, should be in my view, or are likely to become commoditized. While OSS has become very popular to the point of mutual friends and acquaintances claiming that it’s eating the entire industry, it also has problems. First and foremost the very nature of the model doesn’t provide any differentiation so it is primarily a cost reduction model, though also has had a big impact on interoperability as well. Most of the world’s economy is dependent upon differentiality, and since technology has become such an influential force in business–deterministic in many sectors, this is not a trivial problem that can be wished away due to service models, ideology, or regional bias. The OSS model is no longer as open as it once was. It has morphed into something more like the sharing economy, which still exploits the heck out of some while favoring others, and is anything but pure. OSS is often used as job protection inside customer organizations, justifying employee and contractor compensation rather than SW license and fees. Fine to a point–particularly IMO for creating competition where it doesn’t exist otherwise, but the model often fails to deliver for the customer mission too. As we all know OSS is embraced by the service model in vendors, but I and others argue that the service model has already grown too large. In fact it’s shrinking in some very important respects, companies and regions, while many of the service companies are feverishly attempting to improve their IP, IC, and proprietary tech. Both margins and growth are a problem.Now what’s truly fascinating to me about how this has evolved is that some of the dominant proprietary systems have become old, outdated, very expensive and ubiquitous, so they share the negatives of OSS but few of the positives. If most of your competitors are using the same systems from a technical market perspective it no longer offers an advantage, but rather in some cases may be a tax as cost of entry. That works just as long as no alternatives exist or the cost becomes so high that it no longer matters. I think a fair portion of the industry has reached that tipping point.Finally, there is no law that says proprietary needs to be expensive or OSS cheap. Indeed increasingly is the case where as they morph exceptions such exceptions have become more common. That is to say that it is possible with the right design and strategy to have your cake and eat it too, while mitigating or eliminating the significant potential time bomb of IP liability many orgs face.

      1. says:

        PS – used numbers for clarity but apparently doesn’t like my Safari browser or v/v… 

    2. says:

      Hi Den, if you think that HANA runs through Eclipse and in the AWS cloud (little cost if spot pricing is used) then your HANA instance is already open source as you can build your own tables and views in it. S/4HANA is a different story. it potentially stores many public companies’ financials and is perfected every day at a very considerable cost to the company. SAP won’t prevent you from developing your own artifacts to process those financials, but you won’t get any support and i’m sure your auditors will have none of that, either. i haven’t seen IBM’s mainframe code joining the open source movement for the simple fact, that, their financials, too, run on it, despite being one of the largest SAP customers. so, there’s linux, but there’s also aix, there’s hadoop (or spark) and there’s db2, and there’s apache and there’s websphere. so, free license and the not so free license live alongside each other just like they always have. thx, greg

      1. says:

        Greg,That is not the same thing, the SAP ABAP code base is Source Open, which means you can update the code yourself and go through the painful process to have your code merged into the main codeline if you want (I can think of very few occasions this happens though)What Dennis is talking about is the actual database code itself being OpenSource, the same way Cassandra, Riak and MariaDB are OpenSource with the community directly contributing to the codebase and adding features.You do hit on a key point though Greg – OpenSource does not mean free. There are usually free versions available, like with RHEL or Suse developer/workstation editions but Enterprise server editions are paid for. There is little reason why this would not work for HANA, As Thomas above said, the HANA Developer edition is already free which is a good step.I would love to see this happening, it would create a great opportunity for SAP to be open to a much wider world and developer ecosystem. Their startup program has shown the achievements that can be made using SAP technology and support – imagine if the SAP insiders who are fearful of this move embraced it and the wider community, how many markets would open up. Imagine if the wider community didn’t look at SAP like their uncool fathers with amazing toys but embraced the experience and used the toys to do something amazing instead of reinventing the wheel every time. Chris

    3. says:

      HANA Developer Edition is a free license which allows full development see options here

      1. says:

        I’m sure it’s not intentional and I get the point Thomas but I think we both know that free developer licensing is not the same as open source.