Hey SAP - stop flirting with Cloud Foundry and propose

Dick Hirsch Profile picture for user rhirsch October 4, 2014
Dick Hirsch makes the radical case for SAP getting much closer to Cloud Foundry. In his view, the business case is obvious.

cloud foundry
Recent events started me thinking about SAP’s HANA Cloud Platform (HCP) and its future. For those unfamiliar with HCP, this is the platform through which SA expects developers to build extensions to its own solutions. In this story, I am proposing a much closer linkage between SAP and Cloud Foundry. I'd like to see a marriage. Let's check off those events:

1. Google cloud offered $100,000 credit for start-ups moving to their cloud

Google is serious about winning over startups. They are, after all, the heart and soul of the AWS constituency, and Amazon has wielded cloud credits very productively to keep startups aboard and coming back for more services. Last week Google ripped a page out of that playbook, announcing  $100,000 in Google Cloud credits to qualified startups. That ups the ante over AWS, which offers $25,000 in cloud credits under similar terms. [SOURCE]

2. Cloudbees discontinued their public PaaS

Cloudbees recently announced that they would be discontinuing their public PaaS (RUN@cloud) to focus on their Enterprise Jenkins solutions:

Taken together, these changes demonstrate market consolidation, platform commoditization, a continued strength of on-premise solutions in the enterprise, and the important strategic leverage to be obtained by combining IaaS, PaaS and managed service offerings. Longer term, it calls into question whether there will even be a PaaS marketplace that is identifiable except by the most academic of distinctions. [SOURCE]

The background story about this decision makes very interesting reading and I’d suggest reading it yourself for details behind these general trends.

3. Oracle pushed its PaaS at OOW

While Oracle has been building out the various pieces of its cloud portfolio steadily over the past couple of years, it seemed like Ellison saw this year’s conference as a chance to tell the industry that the entire meal is now fully baked.

One of the main courses is Oracle’s entry into the PaaS (platform as a service) market, with which customers can move on-premises Java applications to Oracle’s cloud database and WebLogic server cloud with “a push of a button,” Ellison said. Non-Java applications have a home up in Oracle’s skies too, as they can run on its IaaS, also with just a push of a button.

Oracle’s PaaS also “endows the applications you build with modernity,” according to Ellison, who cited its additional services for social, mobility, analytics and identity management. [SOURCE]

A time for reflection

As I thought about the trends indicated by these events, I brainstormed about the broader impact on HCP.

  • The offers made by Google and Microsoft Bizspark to start-ups was not and perhaps could not be matched by HCP or SAP’s Start-up Focus program. I knew that there was a partner program for HCP indeed, there have been various incarnations such efforts in the past but this program required that partners pay a fee – however small - to gain access to certain benefits such as better access. Compare SAP’s attitude that people should pay to use the platform to that of Google / Microsoft where Google was paying start-ups to use the platform and you realize that a different approach was evident in SAP’s efforts to attract start-ups. Some might say that I am comparing start-ups to partners and that is unfair. Although partially true, it is the attitude towards the ecosystem that is more important. If monetary inducements to increase adoption aren’t planned, other measures are necessary.
  • The fact that CloudBees was discontinuing their public PaaS was distressing and I saw it as a broader malaise in the PaaS market. I considered whether SAP should follow the same strategy and give up on HCP.

How might SAP deal with such challenges?

Move closer to Cloud Foundry

SAP has been flirting with Cloud Foundry (CF) for a while sponsoring the Cloud Foundry Foundation , developing a service broker for HANA for CF. Yet, a tighter integration has been slow in coming. A recent analysis by Holger Mueller reveals that a CF runtime in HCP might finally be available in the near future. See image below:

CF-HCP planned innovations

As I considered this new move, I thought about the specific motivations behind the change.

SAP could transition from an aging foundation (take a look at the virgo-dev mailing list to measure the degree of community activity) to a more modern infrastructure that has broader community support and a better chance at becoming the PaaS standard.

Yet, the average HCP developer has very little direct involvement with the underlying technical layer where CF would be present and where today the Virgo-based infrastructure exists. They might use the command-line tool to deploy / administer their applications but everything else is largely hidden by different UIs ( Administrator , etc). Indeed, that is the whole idea behind a PaaS – the developer doesn’t have to deal with the lower layers of the platform. Therefore, a change to CF might have little immediate benefit for such developers.

Yes, the use of CF would permit the use of build-packs that are not currently available in HCP and Docker. In a world where many enterprise customers are just starting on their cloud journey, the primary target developer audience for HCP – associated with the existing SAP customer base – would probably not able to fully exploit the potential of a container-based development paradigm that a move to Dockerized HCP would represent.

Note: SAP is involved in the Cloud Foundry development associated with Diego and Docker. Momchil Atansov from SAP was involved in the initial Diego project in the cloud-foundry-incubator project.

Could there be other reasons behind a move towards CF?

A more important motivation might relate to the entity performing the platform / PaaS operations. For the existing incarnation of HCP, the entity benefiting here would be SAP itself since it operates HCP as a public PaaS. Yet could the future of HCP lie along a different path?

Broadening the usage patterns

A recent story by James Watters VP of Product at Cloud Foundry entitled The Great Divide: Reflections on the Public/Private PaaS Market in 2014 examines the public cloud market:

In contrast to the startup failures in the public PaaS market, the revenue in the private platform market is significant and growing exponentially. While many enterprises are leveraging public cloud, many of the “crown jewels” applications still reside in private data centers, along with the majority of IT budgets. This means that despite public cloud adoption, architectural standards often start in the private cloud landscape and extend out.

I pinged James after reading an early version of the blog to understand the implications for the dominant enterprise software vendors:

As I thought about his response, I realized that pushing HCP into private clouds would require that partners and customers understand how to operate HCP in such settings. Since CF is gaining ground in a variety of environments, the likelihood of related operations experience and the willingness to gain such knowledge - if not already present - was higher if CF was the foundation of HCP.

Indeed, in his response to a story by Phil Wainwright, HCP evangelist Matthias Steiner confirmed that this was one of the intentions behind the move towards CF.

One more thing: as announced a while back, SAP joined the Cloud Foundry Foundation and there are concrete activities happening right now to incorporate Cloud Foundry as an additional runtime container into HCP. One of many reasons for this decision was to gain more flexible deployment models, e.g. running HCP in non-SAP data centers and private cloud scenarios.)[ SOURCE ]

The ability to host HCP in such environments would expand beyond the traditional SAP ecosystem to include CF partners as well.

The HANA Enterprise Cloud Angle

If we look at moving HCP into private clouds, then one interesting possibility is running the PaaS in SAP’s HANA Enterprise Cloud (HEC), its own managed cloud offering.

Note: There are other examples of other CF-based enterprise PaaS solutions taking the same route. For example, HP Helion Managed Private Cloud is a CF-based PaaS that runs in HP’s data centers.

I already discussed the potential advantages of this union at SCN in 2013 with the title: The first heresy: Could Cloud Foundry run on the HANA Enterprise Cloud?

As I’ve stated before, the HEC gains in value if extensions to the hosted applications can be hosted in a single environment rather than multiple platforms. Thus, the combination of a PaaS and the HEC might be valuable to customers.

There are certain hints, a tweet from Global Vice President of Strategy for Cloud Solutions Sven Denecken and a tweet from HCP Evangelist Mathias Steiner that suggest SAP is examining such options.

The competition doesn’t sleep

All of this is powered by an underlying technology infrastructure that is getting close to achieving what some may have considered an impossible goal. Workday is evolving towards an enterprise-grade transactional system that runs on a best-in-class consumer Internet application architecture.

In contrast to the single-point-of-failure, scale-up designs of traditional enterprise software Workday is evolving towards a  Netflix-like microservices-based architecture, in which every microservice is independently owned and operated by a separate development team. [SOURCE]

As I read Phil's recent story, I was struck by the use of microservices in Workday’s architecture. Such usage patterns are usually associated with consumer applications as well as CF . Here was an excellent example of an enterprise software vendor bridging both enterprise and consumer worlds to attain a competitive advantage.

Another competitor: Oracle is aggressively attacking the PaaS market. More worrisome for HCP was the partial overlap of target audience – existing customers moving their applications towards the cloud. This focus was prominent during Ellison’s OOW PaaS-related demos.

Ellison wanted to show that it's possible to "modernize and lower the cost" of on-premises applications by moving them into Oracle's cloud "with a single click." [SOURCE]

Propose to Cloud Foundry

In the face of competitive threats at both the SaaS and PaaS level, how can SAP respond? A tighter association with Cloud Foundry would provide numerous benefits, especially versus non-CF-based PaaS competitors. A CF-based HCP could provide SAP with the technical foundation to create similar architectures for its SaaS solutions and those applications created by the broader SAP ecosystem as that proudly exhibited by Workday.

While the Oracle PaaS primarily targets scenarios enabling the transition of legacy applications to the cloud, a CF-based HCP could promote additional scenarios such as IoT and mobile more effectively than it already does and help accelerate developer adoption regardless from which ecosystem they originate but who are interested in more modern architectures such as microservices.

Furthermore, a tighter union with CF would enable other hosting partners to promote HCP in private clouds. There are numerous other CF-based PaaS offers such as IBM BlueMix and HP Helion yet none has moved aggressively in this direction. A rapid expansion in this market segment would provide HCP a long-term advantage and allow it to capture this market early in the business cycle.

In short, SAP should stop flirting with Cloud Foundry, bend a knee and propose – now.

A grey colored placeholder image