Skip to Content
Author's profile photo Jan Penninkhof

When the River runs dry

/wp-content/uploads/2012/07/dry_riverbed_118105.jpgIt has been awfully quiet around the River platform. After its initial announcement in 2010, private mentor program, beta programs, trials and innojams, the service apparently didn’t catch enough steam for a successful birth. Instead, SAP management has decided to discontinue SAP River platform development, with the exception of SAP Carbon Impact. To me, this came as quite an unexpected move, as most reviews and blogs on SAP River showed a very positive verdict.

Reasons for shutting down River

One can only guess about the reasons for the shutdown:

  • River is a RAD tool based on data modeling. Is a modeling tool perhaps not the preferred development environment of cloud developers? In a private SAP Mentors meeting Hasso once pointed out that modeling does not mean much for developers.
  • Was it too data-centric? If business logic needed to be added, developers often needed to fall back to the not-so-RAD graphical Action editor (that I somehow always associated with the MIT developed programming language “Scratch”). Actions could be implemented using JavaScript though (Codebox), which made it very powerful in my opinion. From these codeboxes it was possible to call external systems and existing business rules (e.g. service-enabled rules in BRF+) could be re-used.
  • Did it still take too much effort to build applications? Although the data-model could be setup very quickly, adding business logic, applying a custom skin to an application or understanding River’s REST API had a huge learning curve to my experience. And if relatively complex application like SAP’s Carbon Impact needs to be built, it would probably still require a little army of developers.
  • Has SAP not been clear enough in its positioning of River and what it could be used for?
  • Although it’s all about “Run Better” at the moment, many haven’t forgotten about the “Run Anywhere” yet. This doesn’t only apply to the plethora of devices modern software is supposed to run on, but to my opinion this should also apply to cloud vs on-premise. Is it because building an on-premise version of SAP River was too challenging and did customers/partners find the risks of running on a cloud-based platform only too high?
  • Is it expected that the RAD tools built in SAP River be surpassed by Eclipse based RAD-like tooling? SAP has proven to be good at building Eclipse based modeling tools, such as the toolset built for Web Dynpro, the Composite Application Framework (CAF), but also the initial code of the JPA Diagram Editor (which was awarded Eclipse Best Developer Tool award for 2011). And I can’t wait for the expected goodness around Neo, ABAP in Eclipse and SAPUI5.

I tend to believe that the real reason is probably a mix of some of the points mentioned, although I strongly feel that the latter point is most crucial. SAP is betting on Netweaver Cloud (Neo) and together with a little boost in eclipse based developer productivity tools, developer productivity could be on par with SAP River’s modeling tools. And if you’re betting heavily on a platform, it makes sense to shift many of your resources there.

The legacy

Several years of development weren’t in vain. I am very sure that internally, many new ideas have sparked of from the River project:

  • The River project must have gained SAP a lot of insight in the technology that it currently applies. River was their first PAAS and to some extent, you could even call it the predecessor of SAP Netweaver Cloud. SAP River was also one of the first adopters of SAPUI5, and applying it to a prestigious platform as River would probably provide the SAPUI5 team with a lot of feedback. SAP River was also on of the first platforms that showed a REST interface, allowing for a more loosely coupled user interface layer.
  • RDL (River Definition Language) started off as a new language that was targeted to be embedded in River. It has eventually been embedded within the HANA server as the application language, to build native HANA applications.

The future

It has taken approximately five years to introduce HANA as a product (2010) after SAP’s initial acquisition of Transact In Memory (2005). It wouldn’t surprise me if a few years down the road, we would see a new announcement from SAP, starting with “Remember River of back-then?…”

Assigned Tags

      You must be Logged on to comment or reply to a post.
      Author's profile photo Jim Spath
      Jim Spath

      Another non-starter I'm glad not to have started with.

      Author's profile photo Jan Penninkhof
      Jan Penninkhof
      Blog Post Author

      I think a lot of people will say the same.

      Personally, I don't regret putting some time into it though. It was my first experience to really work with REST interfaces. And I think I would also never have experimented with BI On Demand if I wasn't challenged during Innojam Netherlands, where our team was trying to push River data to BI On Demand. Learned a lot on the way!

      Author's profile photo Fred Verheul
      Fred Verheul

      Hi Jan,

      Thanks for writing this (excellent) blog!

      Like you, I was quite surprised (and saddened) to hear this news, even though the partner instance was shut down earlier this month. The message I got from new product owner Mickey Hoter (in early May) was: "The official current SAP decision is that River should be for internal use, meaning it will not go officially to the market for the time being". This sounds less definitive than your statement above. I'm wondering if we should've shouted earlier and louder to maybe reverse this decision...

      I also remember a quote from Vishal Sikka from an ASUGNews interview (last paragraph) in which he says "I believe that I will be able to find time to get River right, and it will revolutionize the way we build applications. I think that will be a bigger thing than HANA and that is a natural counterpart to HANA." I would like to know if he's changed his mind meanwhile, or whether he just hasn't found the time yet...

      Nice pic by the way 😉 .

      Author's profile photo Jan Penninkhof
      Jan Penninkhof
      Blog Post Author

      Thanks Fred. I totally understand where you're coming from. It was quite obvious to me, that you have put a lot of time and effort into understanding the system and building applications on top of it. Your River based application at Innojam last year was absolutely stunning from both a technical as well as business perspective, and I always hoped that it would eventually see the light to help troubled kids.

      In my opinion, the types of applications we built, was exactly what River was really good at. Rapidly building data driven applications for environments that didn't really care about a totally thought-through user-interfaces and for people that needed a system fast, and could live with a few short-comings.

      It is quite interesting that most of the applications we built came down to applications for charity organisations. Of course the Innojam theme contributed to that, but getting actually working applications out of Innojam that could really be deployed at those charities exceeded my wildest imagination. When Benjamin Wesson told me during Teched last year, that he also built an application for a food bank in the US using River, it convinced me that we were not the only ones that found a perfect niche for SAP River.

      And yes, perhaps we should have pushed harder from the community to keep River alive. But on the other hand, I do understand and respect SAP's decision to discontinue the platform. As mentioned in my blog, I believe SAP has refocussed on SAP Netweaver cloud (Neo) and put their resources there too. And I'm pretty sure that, as time comes, we'll see River modeling tools in the rebound. Perhaps in the Neo SDK, or maybe even in the revolutionary next big thing Vishal Sikka was talking about.

      Author's profile photo Matthias Steiner
      Matthias Steiner

      Nice blog... got little to add.

      Just wanted to emphasize on the last two paragraphs. While I wouldn't go so far to call River a PaaS, nor call it a successor of SAP NetWeaver Cloud, I do like the fact that you highlight the River's accomplishments and the role it played to get to this point. And I like that you hint that the technology is far from running dry 😉

      Author's profile photo Kumud Singh
      Kumud Singh

      And..... 'When the River runs dry' what happens to the fishes? 😉 Did they find an ocean?



      Author's profile photo Matthias Steiner
      Matthias Steiner

      Fortunately, the number of fishes should be small and they have all be alarmed before hand, so losses should be minimal. 😉

      Author's profile photo Jan Penninkhof
      Jan Penninkhof
      Blog Post Author

      So Long, and Thanks for All the Fish 😉

      Author's profile photo Bjoern Goerke
      Bjoern Goerke

      Hi Jan,

      thanks for this great blog post.

      Actually, many reasons you list somewhat apply, but from my perspective the main considerations have been the following:

      1. Since the "original" River project started, many things have significantly taken shape within SAP that did not exist at the time, namely HANA and SAP NetWeaver Cloud (aka Neo) in this case

      2. We're meanwhile developing project River into something bigger than what it was before, making specific adjustments in the concepts to leverage HANA; you mentioned RDL (River Definition Language) which is the core part of this effort spanning "runs anywhere" in the sense that parts of RDL will be available in "River" (as a RAD environment) but also in ABAP and Java and most importantly in HANA natively

      3. With SAP NetWeaver Cloud as an open and standards-based environment, any JVM based RAD environment can be used to create applications efficiently (take SpringRoo, JRuby or others as an example); as you mention above, Eclipse is our preferred -- but through the Neo SDK not exclusive -- development environment

      4. Plans are, based on RDL, to provide a successor to "River back-then" providing RAD-like capabilities

      So the river's not really running dry. Only running low perhaps right now to an external observer. But chances are it'll bring back a spring's flood some time soon... 😉 Stay tuned...

      Best regards,

      Björn Goerke (Technology & Innovation Platform Core)

      P.S.: All statements with the usual disclaimers: Any recourse to courts of law is excluded.

      Author's profile photo Jan Penninkhof
      Jan Penninkhof
      Blog Post Author

      Thank you very much for your reply. With your internal views and insights into this matter, I think this blog post has become multiple times more valuable.

      Thanks also for the even further clarifying, motivational and maybe even poetic tweets:

      • "I think the most interesting aspect is that there will be a new environment with RAD-like features"
      • "Relationship between RDL and new RAD environment is still unknown"
      • "Like with all rivers: while the name stays, the scenery changes as you follow its flow ;-)"

      Jurgen Schmerder replied with a very nice tweet as well, summarizing its course with:

      • "Product will die, vision lives on"