Open Source Strategy

CIOs can generate a sustainable business value with OpenStack! Here’s how it works.

OpenStack is on everyone’s lips and is slowly working it’s way forward into the IT infrastructure of German companies. 58 percent of German CIOs see in OpenStack a true alternative to commercial cloud management solutions. However, IT decision makers should look closely what they are holding in their hands. OpenStack is basically just an infrastructure management solution and accomplishes no direct value to the business success. However, this is exactly what CIOs have to deal with in order to position themselves as a business enabler. This raises the question how OpenStack is able to create a significant added value, stepping out of the shade of a simple open source cloud management software.

CIOs have to state the fundamental question, how OpenStack can provide a strategic advantage. This is only the case if they use OpenStack different from their competition and thus not only restrict oneself to operational excellence. It is rather about to understand the OpenStack technology as part of their IT strategy using it to create a real value for the company.

Playground: Enablement Platform for Developers

CIOs who want to achieve a business value with OpenStack need to see more potential in OpenStack than just a pure management solution for their cloud infrastructure. In this case it is much more than just cost savings and running an IT infrastructure. CIOs have to understand OpenStack as an enablement platform for their developers and use the open source solution exactly in this way. CIOs, who only want to focus on operational excellence, can use one of the several standard OpenStack distributions. However, these ones should consider, that they are just one of many. For those, who see a strategic vehicle in OpenStack, a distribution is a good foundation to extend OpenStack and expand it to an enablement platform.

OpenStack as enablement platform means that developers are provided with much more than just compute (virtual machines), storage or databases. It is about providing higher-value services, which can be found in the portfolios of Amazon Web Services and Microsoft Azure. This is about microservices that support the application development. These are ready building blocks whose functionalities don’t have to be developed again. Instead they can be used directly as a “brick” within the own source code. The OpenStack community has recognized the importance and slowly tries to take parts of the AWS respectively Azure service portfolio over to OpenStack. First services are Sahara (Elastic Map Reduce) and Zaqar (Multiple Tenant Cloud Messaging). Of course, this is by far not enough and further microservices are needed to make OpenStack to a powerful enablement platform.

However, one thing shouldn’t be underestimated for this scenario: the significant investments that are necessary. Building a massive scalable and globally available cloud infrastructure respectively enablement platform like the ones of Amazon Web Services or Microsoft Azure is complex as well as costly. However, this must not prevent CIOs from running an own OpenStack based platform. For this purpose e.g. partnerships with hosting providers suit to run OpenStack in different deployment models. The German research paper “Managing OpenStack: Heimwerker vs. Smarte Cloudsourcer” describes what kinds of variants are suitable.

Despite the high complexity of OpenStack it is possible for CIOs to make use of its openness and flexibility to build custom solutions based on the various sub-projects.

Innovation: Leave the Community

The common work within a community is important to push a project like OpenStack successfully forward. In addition, all involved parties benefit from the ideas other community members. However, the big disadvantage is that one member is just as good as the community itself. The community concept neither works to diversify towards the competition. In the end the focus is again completely on operational excellence. A technological or strategic advantage cannot be achieved.

The cloud market shows that solo efforts are part of a concept for success. Providers like Amazon Web Services (AWS), Microsoft Azure, ProfitBricks in Germany or CloudSigma in Switzerland have built individual infrastructure environments. This strategy has helped AWS to achieve a massive advantage in technology within the cloud market.

Another paragon in the open source space is Canonical. The Linux distributor is well-known for using the Linux open source code but only giving back a little (e.g. patches). It is the same behavior for the OpenStack project, proven by numbers (see Idealists my have a problem to understand Mark Shuttleworth’s (Founder of Canonical) attitude. However, in the end it is about the business and thus to operate in the black.

OpenStack is perfect for CIOs to serve as a foundation for their cloud infrastructure. For this purpose, standard OpenStack services like compute, storage or identity management can be used. However, in the end it is about one thing: individuality! This means that a CIO should say good-bye to the community concept in order to focus on innovation. The strategy is about to look what the OpenStack community has to offer, adopt the necessary services and develop individual enablement services and solutions on top of it. The use of standard OpenStack is not sufficient to differentiate from the competition neither to create a serious value for the company.

Business Value: Applications and Services support new Business Models

The Internet of Things (IoT) is the next megatrend rolling towards companies forcing them to finally start with their individual digital transformation. OpenStack is a powerful tool and a global standard CIOs can use to support this individuality. However, OpenStack is just a means to an end and only offers the basic functionality so far. But, based on the open source approach OpenStack can be completely customized for the own necessities and thus is the perfect foundation for individual backend solutions to support mobile and IoT applications.

For this purpose, CIOs should stand back from standard OpenStack, use a distribution as a foundation and extend it with individual functionalities. This means that they have to get used to OpenStack’s complexity. However, it is exactly this technological excellence that will pay off over time to differentiate from the competition in order to obtain a technical advantage and thus create an added value.

Open Source

OpenStack Acquisitions: The sellout runs at full speed

This year OpenStack celebrates its fifth birthday. Despite its young past the open source project already caused quite a stir, driven by a solvent ecosystem consisting of the who is who of the cloud and open source industry. This has a positive affect on the market maturity. Linux needed around 10 years and thus twice as long to achieve similar market relevance. The OpenStack Foundation has learned from the older brother and does a lot of things right. One signal for the importance and market maturity of OpenStack is the acquisition behavior. The sell out runs at full speed.

OpenStack acquisitions over time

After only five years, the OpenStack community already looks back at a proud number of acquisitions. The year 2014 can be regarded as the starting shot when the first big IT vendors started their shopping tour to snap the strategic most important OpenStack startups for their portfolios.

  • April 2014: Red Hat buys Ceph storage vendor Inktank and transfers the product as „Red Hat Ceph Storage“ in the own portfolio.
  • Juni 2014: Red Hat buys OpenStack integration service provider eNovance to increase its consulting competency for its customers.
  • September 2014: Cisco buys OpenStack-as-a-Service provider Metacloud and transfers the product as „Cisco OpenStack Private Cloud“ in the own portfolio.
  • Oktober 2014: EMC buys the OpenStack distribution Cloudscaling and will release an own OpenStack product called „Caspsian“ in November 2015 that is based on the Cloudscaling technology.
  • Juni 2015: Cisco buys OpenStack pioneer Piston Cloud Computing to strengthen its Intercloud initiative.
  • Juni 2015: IBM buys OpenStack private cloud provider BlueBox to expand its hybrid cloud offering.

During this wave of mergers the big vendors have acquired most of the hitherto most promising OpenStack startups. This shows that OpenStack has made it on top of the strategic agendas of the big players and underwrites its importance. The vendors expand their portfolios but also want to become an inherent part of the OpenStack market. In addition, they buy todays rare OpenStack knowledge from the market.

However, one attractive target is still not taken: Mirantis. Even if the self-appointed pure play OpenStack vendor is today’s highly profiled acquisition candidate. With good reason, Mirantis is offering an own distribution (Mirantis OpenStack), a hosted private cloud (Mirantis OpenStack Express) as well as consulting services (Mirantis Consulting) and training services (Mirantis Training). So, a quite appealing target for the prospective buyer to make the portfolio OpenStack ready in all aspects and that at a single blow.

The shopping tour of the big vendors speaks for their commitment and the progressed maturity stage of OpenStack. Indeed, a good sign for the further development of the open source project. However, it seems that the OpenStack ecosystem more and more is getting under the control of the big players.

Cloud Computing Open Source

Round 11: OpenStack Kilo

The OpenStack community hits round 11. Last week the newest OpenStack release “Kilo” was announced – with remarkable numbers. Almost 1.500 developers and 169 organizations contributed source code, patches etc. Top supporting companies to OpenStack Kilo include Red Hat, HP, IBM, Mirantis, Rackspace, Yahoo!, NEC, Huawei and SUSE. OpenStack Kilo is characterized by a better interoperability for external drivers, supporting new technologies like container as well as bare-metal concepts.

OpenStack Kilo: New Functions

According to the OpenStack Foundation, almost half of all OpenStack deployments (46 percent) are production environments. Network function virtualization (NFV), for using single virtual network components, is the fastest-growing use case for OpenStack. One of the lighthouse projects is eBay, operating OpenStack at large scale.

Essential new functions of OpenStack Kilo

  • OpenStack Kilo is the first release that fully supports the bare-metal service “Ironic” to run workloads directly on physical machines.
  • The OpenStack object storage service “Swift” supports “Erasure Coding (EC)” to fragment data and store it at distributed locations.
  • The “Keystone” identity service was enhanced with identity federation to support hybrid and multi-cloud scenarios.

New features of the OpenStack Core Projects (excerpts)

  • OpenStack Nova Compute
    Improvements for live updates when a database schema is changed and support the change of resources of a running virtual machine.
  • OpenStack Swift Object Storage
    Support of “Erasure Coding”. Temporary access to objects via an URL and improvements for global cluster replication.
  • OpenStack Cinder Block Storage
    Enhancement to attach a volume to multiple virtual machines to implement high-availability and migration scenarios.
  • OpenStack Neutron Networking
    Extension of network function virtualization (NFV) like port security for OpenVSwitch and VLAN transparency.
  • OpenStack Ironic Bare-Metal
    Ironic supports existing virtual machine workloads as well as new technologies like container (Docker), PaaS and NFV.
  • OpenStack Keystone Identity Service
    The extensions around identity federation help to distribute workloads across public and private clouds to build OpenStack based hybrid and multi-cloud environments.

OpenStack Kilo: Short Analysis and Impact

OpenStack is still growing. Even if a high ratio of NFV use cases shows that OpenStack is mainly used in service provider networks to operate single network components more flexible and cost-effective. However, the new Kilo functions for “federated identity”, “erasure coding” and “bare-metal” will move OpenStack up to the top of the CIO agenda.

The support of “erasure coding” is a long overdue function for Swift Object Storage – even though initial discussions already started for the “Havanna” release in 2013. All big public cloud providers are working with this distribution strategy for years to ensure high-availability of data. The introduction of bare-metal is at the right time. Workload migrations to cloud based infrastructure show with increasing frequency that virtual machines are not suitable for all use cases. Thus, database servers and performance intense workloads are ideally running on physical machines, whereas distributed workloads like application and web servers are good candidates for virtual machines. On a final note, identity federation will help CIOs building seamless OpenStack based hybrid and multi-cloud environments. Users only need a single login to authorize across multiple providers and get access to servers, data and applications in private and public clouds at once.

This begs the question how easy and fast CIOs can benefit from these new functions. The last five years unveiled that using OpenStack implicates a high complexity. This is mainly because OpenStack is organized as a big project composed of several sub-projects. Only the close interaction of all necessary sub-projects to support a specific use case is promising. The majority of CIOs who are working with OpenStack are considering a professional distribution instead of building an own OpenStack version based on the source code of the community trunk. In Germany these are 75 percent of the OpenStack users.

Open Source

Open Source and OpenStack: Complexity and lack of knowledge raise risks

Open source solutions offer a cost advantage with their free licenses or low license costs. However, there are costs associated with these solutions that should not be neglected. On the one hand, specialized knowledge is necessary to build and develop open source cloud infrastructure. On the other hand, administrators have to ensure proper infrastructure operations, for which task extensive skills regarding solution administration and maintenance are requisite. In most cases, these skills are acquired via external expert knowledge such as advisory or developer resources.

Furthermore, users who decide to build their cloud on pure open source software, are limited to the support dependency of the open source project. This could be tough and painful, since support is generally provided by forums, chats, Q&A systems and bug tracking systems. In addition, it is well received when users are actively participating and playing a part in contributing to the project, a behavior generally not needed in the commercial world of software. Luckily, commercial-oriented vendors of open source cloud software have already identified the service gap and provide support through professional services, as part of special license plans for enterprises.

Despite a high level of flexibility and openness as well as a decrease in license costs, it is inherently important to understand that OpenStack is not just a piece of software. Instead, the open source cloud management solution is based on a collection of specific components, among others, for compute (Nova), storage (Swift, Cinder) and networking (Neutron) capabilities, which must be tightly integrated to build a complete and powerful OpenStack based cloud environment. The whole project takes in new functionalities with each release. In addition, a community of developers and vendors participates with further add-ons and source code for maintenance purposes and other improvements. Therefore, the use of OpenStack can lead to unpredictable complexity and overall risk increases.

IT organizations who attempt to deal with this complexity on their own by integrating all components from scratch and being up-to-date at all times, tend to expose themselves to the risk of creating their own, unmanageable cloud solution instead of using an industry compliant standard. The precise customization of OpenStack to the individual company requirements can easily lead to an OpenStack environment that is incompatible with external OpenStack based cloud infrastructure. Thus, the connection of internal and external cloud infrastructure in a hybrid scenario becomes quite tricky.

The increasing relevance of OpenStack as a central technology component within cloud environments leads to a higher demand for specialized consultancy, integration and support services. This market is still in the nascent stage and big IT vendors are currently training and improving their staff knowledge. After all, the present supply of readily trained, skilled and experienced OpenStack administrators, architects and cloud service broker is negligible. CIOs should immediately plan how to build basic skill levels for OpenStack within their IT organizations. Even if specialized service contractors can help during the implementation and operation of OpenStack based clouds, IT architects and managers should still have the main responsibility and know what is happening. OpenStack is not an instant meal that just needs to be warmed up but rather a complex technology platform composed of several individual components whose configuration is rather matching the preparation of a multi-course gourmet dinner. Skills and passion are on the most wanted list.

Cloud Computing Open Source

Signature Project: SAP Monsoon

Until now, SAP didn’t make a strong impression in the cloud. The Business-by-Design disaster or the regular changes in the cloud unit’s leadership are only two examples that reveal the desolate situation of the German flagship corporation from Walldorf in this market segment. At the same time, the powerful user group DSAG attempts a riot. The complexity of SAP’s cloud ERP as well as the lack of HANA business cases are some of the issues. The lack of transparency of prices and licenses as well as a sinking appreciation for the maintenance agreements, since the support doesn’t justify the value of the supporting fees, are leading to uncertainty and irritation on the customer side. To add to this, a historically grown and complex IT infrastructure causes a significant efficiency bottleneck in the operations. However, a promising internal cloud project might set the course for the future if it is thoroughly implemented: Monsoon.

In the course of the years, the internal SAP cloud landscape has evolved into massive but very heterogeneous infrastructure boasting an army of physical and virtual machines, petabytes of RAM and petabytes of cloud storage. New functional requirements, changes in technology as well as a number of M&As have resulted in various technology silos, thus greatly complicating any migration efforts. The new highly diverse technology approaches and a mix of VMware vSphere and XEN/KVM distributed over several datacenters worldwide lead to increasingly high complexity during the SAP’s infrastructure operations and maintenance.

The application lifecycle management is the icing on the cake as the installations; respectively, the upgrades are manual, semi-automated or automated, as determined by the age of the respective cloud. This unstructured environment is by far not an SAP-specific problem, but represents rather the reality in middle to big size cloud infrastructure whose growth has not been controlled over the last years.

„Monsoon“ in focus – a standardized and automated cloud infrastructure stack

Even if SAP is in good company with the challenge, this situation leads to vast disadvantages at the infrastructure, application, development and maintenance layers:

  • The time developers wait for new infrastructure resources is too long, leading to delays in the development and support process.
  • Only entire releases can be rolled out, a stumbling block which results in higher expenditures in the upgrade/ update process.
  • IT operations keep their hands on the IT resources and wait for resource allocation approvals by the responsible instances. This affects work performance und leads to poor efficiency.
  • A variety of individual solutions make a largely standardized infrastructure landscape impossible und lead to poor scalability.
  • Technology silos distribute the necessary knowledge across too many heads and exacerbate the difficulties in collaboration during the troubleshooting and optimization of the infrastructure.

SAP addresses these challenges proactively with its project “Monsoon”. Under the command of Jens Fuchs, VP Cloud Platform Services Cloud Infrastructure and Delivery, the various heterogeneous cloud environments are intended to become a single homogeneous cloud infrastructure, which should be extended to all SAP datacenters worldwide. Harmonized cloud architecture, widely supported uniform IaaS management, as well as an automated end-to-end application lifecycle management form the foundation of the “One Cloud”.

As a start, SAP will improve the situation of its in-house developers. The foundation of a more efficient development process is laid out upon standardized infrastructure, streamlining future customer application deployments. For this purpose, “Monsoon” is implemented in DevOps mode so that development and operations of “Monsoon” is split into two teams who work hand in hand to reach a common goal. Developers are getting access to required standardized and on-demand IT resources (virtual machines, developer tools, services) through a self-service portal. Furthermore, this mode enables the introduction of the so called continuous delivery. This means that parts of “Monsoon” have already been implemented and used actively in production while other parts are still in development. After passing through development and testing, components are being directly transferred into the production environment without wait time for a separate release cycle. As a result, innovation growth is fostered.

Open Source and OpenStack are the imperatives

The open source automation solution Chef is the cornerstone of “Monsoon’s” self-service portal, enabling SAP’s developers to deploy and automatically configure the needed infrastructure resources themselves. This also applies to self-developed applications. In general, the “Monsoon” project makes intensive use of open source technologies. In addition to the hypervisors XEN and KVM, other solutions like the container virtualization technology Docker or the platform-as-a-service (PaaS) Cloud Foundry are being utilized.

The anchor of this software-defined infrastructure is OpenStack. The open source project that can be used to build complex and massive scalable cloud computing infrastructure supports IT architects during the orchestration and the management of their cloud environments. Meanwhile, a powerful conglomerate of vendors stand behind the open source solution, trying to position OpenStack and their own services built on OpenStack prominently in the market. Another wave of influence emerges through a range of developers and other interested parties who provide their contributions to the project. At present, around 19,000 individuals from 144 countries participate in OpenStack, signifying that the open source project is also an interest group and a community. The broad support can be verified by a range of service providers and independent software vendors who have developed their services and solutions compatible to the OpenStack APIs. Since its development, OpenStack has continuously evolved into an industry standard and is destined to become the de facto standard for cloud infrastructure.

At the cloud service broker and cloud integration layer, SAP “Monsoon” sets on OpenStack Nova (Compute), Cinder (Block Storage), Neutron (Networking) and Ironic (Bare Metal). OpenStack Ironic facilitates “Monsoon” to deploy physical hosts as easily as virtual machines. Among other things, the cloud service management platform OpenStack is responsible for authentication, metering, as well as billing and orchestration. OpenStack’s infrastructure and automation API helps developers to create their applications for “Monsoon” and deploy them on top. In addition, external APIs like Amazon EC2 can be exploited in order to distribute workloads over several cloud infrastructures (multi cloud).

On the one hand, this open approach gives SAP the ability to build standardized infrastructure, in order to support VMware vSphere alongside OpenStack. On the other hand, it is also possible to execute hybrid deployments for both internal and external customers. The on-demand provisioning for virtual as well as physical hosts completes the hybrid approach. Compared to virtual machines, the higher performance of physical machines shouldn’t be underestimated. HANA will appreciate it.

Study: SAP weigh as a powerful OpenStack partner

SAP’s open source focus on OpenStack is nothing new. First announcements have already been made in July 2014 and show the increasing importance of open source technologies to well-established industry giants.

In the meantime, SAP’s OpenStack engagement also got around on the side of the user. In the context of the very first empirical OpenStack study in the DACH market “OpenStack in the Enterprise”, Crisp Research asked 700+ CIOs about their interests, plans and operational status of OpenStack.

The study concluded that cloud computing has finally arrived in Germany. For 19 percent of the sampled IT decision makers, cloud computing is an inherent part on their IT agenda and the IT production environments. 56 percent of German companies are in the planning or implementation phase and are already using cloud as part of first projects and workloads. One can also say that in 2014 the OpenStack wave has also arrived in Germany. Almost every second cloud user (47 percent) has heard of OpenStack. At present, already 29 percent of the cloud users are actively dealing with the new technology. While 9 percent of the cloud users are still in the information phase, already one in five (19 percent) have started planning and implementing their OpenStack project. However, only two percent of the cloud users are using OpenStack in their production environments. Therefore, OpenStack is only a topic for pioneers.

On the subject of performance ability of OpenStack partners, the study has shown that cloud users supportive of OpenStack appreciate SAPs OpenStack engagement and respectively expect a lot from SAP. Almost half of the sampled IT decision makers attribute “a very strong” performance ability to SAP, IBM and HP.

„Monsoon“ – Implications for SAP and the (internal) Customer

In view of the complexity associated with “Monsoon”, the project rather deserves the name “Mammoth”. To move a tank ship like SAP into calm waters is not an easy task. Encouraging standardization within a very dynamic company will raise anticipated barriers. In particular when further acquisitions are pending, the main challenges are to integrate these into the existing infrastructure. However, “Monsoon” seems to be on the right way to building a foundation for stable and consistent cloud infrastructure operations.

As a start, SAP will benefit organizationally from the project. The company from Walldorf promises its developers time savings of up to 80 percent for the deployment of infrastructure resources. As a consequence, virtualized HANA databases can be provided completely automated, thus decreasing the wait time from about one month to one hour.

In addition to the time advantage, “Monsoon” also helps its developers to focus on their core competencies (software development). In former times, developers were involved in further processes such as configuration and provisioning of the needed infrastructure; now they can independently deploy virtual machines, storage or load balancers in a fully automated way. Besides fostering the adoption of a cost-effective and transparent pay-per-use model where used resources are charged by the hour, standardized infrastructure building blocks also support cost optimization. For this purpose, infrastructure resources are combined into standardized building blocks and provided across the SAP’s worldwide datacenters.

The introduced continuous delivery approach by “Monsoon” is well positioned to gain momentum at SAP. The “Monsoon” cloud platform is regularly extended during operations and SAP is saying good-bye to fixed release cycles.

External customers will benefit from “Monsoon” in the mid-term, as SAP is using collected experiences from the project during the work with its customers and will also flow them into future product deployment (e.g. continuous delivery).

SAP is burning too many executives in the cloud

SAP will not fail with the technical implementation of “Monsoon”. The company is employing too many high-qualified employees who are equipped with the necessary knowledge. However, the ERP giant is incessantly showing signs of weakness on the organizational level. This vehemently raises the question why ambitious employees are never allowed to implement their visions to the end. SAP has burned several of its cloud senior managers (Lars Dalgaard is a prime example). For some reason, committed and talented executives who try to promote something within the company seem to have a tough act to follow.

SAP should start to act according to the terms of its customers. This means not only thinking about the shareholders, but also following a long-term vision (greetings from Amazon’s Jeff Bezos). The “Monsoon” project could be a beginning. Therefore, Jens Fuchs and his team are hopefully allowed to implement the ambitious goal – the internal SAP cloud transformation – successfully to the end.

– – –
Image source: Christian heinze /