Enterprises to make use of High-Performance Computing to
improve their products and services?
A case study of a High-Performance Computing infrastructure and service provider looking to diffuse its HPC technology into a new market segment: Small
and Medium-sized Enterprises.
Author: Zuhair Salem el Husseini
Student ID: 10839615
September 14
th, 2015
Abstract
The quest towards achieving a sustainable competitive advantage is encouraging more companies and institutions to invest in technological processes and systems that enhance their efficiency and quality of services. Yet, at what cost and what challenges and benefits? The purpose of this paper is to explore the benefits and processes of high process computing for private and public institutions, in addition to the shortcomings and barriers towards fully adopting these processes. It also aims to analyze and interpret the adoption of new technology from the point of view of users, before finally identifying the marketing role of the technology service provider. As far as the methodology is concerned both quantitative and qualitative research designs were used, utilizing both credible websites, journal articles in addition to personal interviews with relevant experts for the qualitative research, while also referring to a statistics from a quantitative survey. More specifically, benefits were seen in product differentiation, internal and external communication, research and development, innovation and consequently increased ROI, providing the company with a sustainable competitive advantage. The findings indicate that the advantages of High-Performance Computing (HPC) for businesses of various sizes have proven to be feasible on the long run, enhancing the results in terms of efficiency, quality of products and services, yet there are barriers related to the upfront costs, needed expertise and compatibility of the technology systems with the clients’ needs. The paper also discusses how high performance computing can potentially connect businesses and public institutions with their stakeholders, consumers, partners and their own internal elements in ways that other traditional forms of computing and technology cannot.
Table of Contents
Abstract
2
Acknowledgments
4
I. Introduction
5
II. Conceptual Frameworks
8
A. Technology Acceptance Model
8
B. Resource-based Theory
10
C. Diffusion of Innovation
15
III. Case studies
21
IV. Findings & Analysis
41
V. Conclusion
49
References
51
Acknowledgments
At the outset, I would like to extend my sincere appreciation to Prof. Dr. Anwar Osseyran for granting me the opportunity to work closely with his company, and without whom this work would have been difficult to achieve.
I would also like to thank my supervisor Dr. Jean L Johnson for her valuable assistance throughout the course of this work.
I would like to pledge my sincere gratitude to Chantal Cassee, Frank Heere, and all my colleagues at SURFsara for their considerable support and guidance through various stages of this project.
For their encouragement and support, I thank my family, friends and the MBA office at the Amsterdam Business School.
Zuhair Salem
I. Introduction
In this era, sometimes called Information Age, whilst the activity of human affairs is growing at an increasingly rapid pace, and due to the increasing demand for higher processing speed and faster computing, the leaders of the technology industry and computer architects are working day and night to meet these needs.
Today, the use of information technology by businesses is inevitable and very few dispute its importance. Not only having an information system is now unquestionable, but companies also race up to acquire the latest and most advanced technologies there is to gain competitive advantage (Welsh & White, 2014).
Benefits of technology adoption range from having better information management and internal communication to product development, product differentiation, and innovation (Nguyen, 2009). These benefits generally result in improved overall performance and increased ROI, and hence allowing the company to stand out in the market (Nguyen, 2009).
Emerging technologies widely vary, and the selection of technology tools often depends on the nature of the utilizing business. Among the emerging technologies in the computing domain, High-Performance Computing (HPC) is stealing the focus from other high-tech tools after it proved to be a real game changer (Nguyen, 2009).
High-performance computing generally refer to the practice of aggregating computer power in a way that delivers a much higher performance that one could get out of a typical desktop or workstation in order to solve large problems in science, engineering, or business (Welsh & White, 2014).
Despite its evident capabilities, HPC is rather utilized by governmental agencies, universities and large enterprises, while its presence among small and medium-sized enterprises (SMEs) is still considered to be modest. The market segment that uses HPC is relatively narrow since, in Europe, 99% of businesses are small and medium-sized enterprises (SMEs). What stands between SMEs and their usage of HPC? Are SMEs aware of the capabilities that HPC have and the impact it could invoke onto their performance? How SMEs perceive the usage of HPC? And what is the role of the stakeholders in diffusing technologies?
In this work, we will try to answer these questions by taking a closer look at the recent relationship that has occurred between the HPC technology and SMEs – mostly in Europe – and unveil the reasons for why some industries choose to adopt HPC technology among others, what drives them and what prevents them from utilizing it. We will examine success stories, the enticement that led to adopting this technology and how it helped them improve their overall performance; and on the other hand, we will also take a look at unsuccessful trials of some SMEs to adopt it, and discover the reasons behind their failed attempts (Nguyen, 2009). Moreover, we will explore the efforts exerted by the HPC Service Providers, governmental bodies, and other European partnerships/organizations, in the aim to raise awareness and supply European SMEs with the competency necessary to benefit from the innovation possibilities induced by HPC, and consequently increase their competitiveness.
Using frameworks of Information Systems and Marketing, we will analyze and interpret the adoption of a new technology from a user point of view, through case studies that consist of interviews and surveys with SMEs from different industries that have utilized the technology and that shared
Furthermore, we will define elaborate on the definition of HPC, identify the marketing role of the service provider in the diffusion of HPC onto new market segments, considering the intrinsic cohesiveness of both user and provider perspectives that ultimately lead to the success of the adoption. In addition to a few words from the Managing Director of SURFsara - a prominent personality in the High-Performance Computing community, about the future of HPC in Netherlands and in Europe, and the impact of its diffusion in the economy
Following the case studies, deliverables will be shown in the “Findings & Analysis” section in the form of statistical charts and tables, and then afterwards, we will analyze the results, and draw conclusions which will finally enable us to identify the barriers and incentives for SMEs to acquire HPC technology. Identifying these two elements holds strategic importance to all stakeholders whether they were users, service providers, brokers, and Independent Software Vendors (ISVs), ultimately allowing them to work hand in hand to cross the boundaries and promote the benefits.
At the end, we will wrap it up with a summarized conclusion on the overall case study, and an interpretation of the findings.
II. Conceptual Frameworks
As previously mentioned, joint efforts of Marketing and Information Systems must be carried out in order to optimize a successful targeting campaign of relevant Small and Medium-sized Enterprises. The reason why it must be a collaborative work is that the barriers behind the adoption are technical and promotional in their nature. Even though SURFsara is the sole “HPC infrastructure + support” provider in the Netherlands, marketing efforts still need to be exerted in order to raise awareness among businesses who themselves do not know whether or not this technology can actually be beneficial for them (Osseyran, 2015). Meanwhile, technicians and engineers are consistently working on minimizing the technical difficulties that SMEs face before and during usage.
Obviously, and just like in most business cases, financial obstructions crash the party. However this time, non-monetary solutions will be yielded by service providers and third parties to make sure that SMEs face no more financial difficulties (Welsh & White, 2014).
Before, Lets elaborate a bit on the models and frameworks of Information Systems and Marketing, that the analysis and interpretation of the cases will be based on.
A. Technology Acceptance Model (TAM)
According to Fenech (2015) the TAM model is a theory of information systems which models the usage and acceptance of technological processes and devices by the end users. According to the model, when a
certain technological device or process is used by clients, their decision to use it is mainly affected by two variables:
As per the TAM, developed by Fred Davis in 1989, the two variables that determine the user acceptance of a technology are:
* Perceived Usefulness (PU): Defined as “the degree to which a person believes that using a particular system would enhance his or her job performance". (Fenech, 2015)
* Perceived Ease of Use (PE): “the degree to which a user expects the target system to be free of effort”. (Fenech, 2015)
This theory is manifested in Figure 1 here below.
Technology Acceptance Model (TAM), version 1.
The Technology Acceptance Model is a theory considered to be an extension of the Theory of Reason Action proposed by Ajzen and Fishbein (1975 & 1980) that was derived from the social psychology setting.
As shown in Figure 1, the model comprises 4 stages through which a user’s decision to use evolves.
According to David, certain external variables like situational constraints and individual abilities influence the use of technology indirectly through their effect on the perceived ease-of-use and perceived usefulness. Both of these elements have an impact on the attitude of the user towards technology, which also has an impact on the intention of the users to choose and use the particular technological process
In addition, according to studies, the system’s perceived usefulness is a strong indicator of how the system will be used. For example, there was a significant correlation found between the self-reported usage and the perceived usefulness (r=0, 63), in addition to its self-predicted future usage (r=0, 85). At the same time, the perceived ease-of-use proved to be a dependable predictor of the acceptance of a user of an application or a system and was positively correlated with its current usage (r=45) and its future usage (r=59). Nonetheless, the ease-of-use significantly depends on the testing of the system or the actual practice that eventually takes place.
Nonetheless, the ease-of-use significantly depends on the testing of the system or the actual practice that eventually takes place.
B. Resource-based Theory
According to Kozlenkova (2014) the “The resource-based view (RBV)” model is one that considers the key to a firm’s superior performance that allows it to sustain its competitive advantage is its resources. The approach, which evolved in the 1990s and 1980s, evolved on the basis that organizations should always look inside the company to find its competitive advantages, instead of looking for a competitive external
environment or market, according to Barney (1991), the author of “Firm resources and sustained competitive advantage”. The model below explains RBV and emphasizes its key points:
Resource-based view model
Yet in order for companies to improve their resource-based competitive advantage, Kozlenkova (2014) explains that instead of trying to acquire new resources and skills for every opportunity that arises, it is more feasible for companies to exploit external opportunities using existing resources in a new way. Within the context of the company’s information systems, it should, for example, utilize the spread of cloud computing among individuals and companies to provide firm-level and consumer-level access to its services and e-products through the cloud (Prout, 2015). It can also utilize the availability of statistical and research providers to produce infographics and visualize statistical data and make it available through its information systems to both the consumer and the staff member (Prout, 2015).
Moreover, in the model of RBV, the resources have the major role of assisting organizations to achieve the maximum organizational performance. According to Amit & Schoemaker (1993) there are two kinds of resources: intangible and tangible.
Intangible resources: are those that have no physical presence but can still be owned by the organization (Thong, 2001). This can include, but not limited to: trademarks, brand reputation, intellectual property, image, good will among others. Unlike the physical resources, these are built over time and cannot be simply bought by companies from a certain market. Intangible assets and resources stay within a company and are considered its main source of sustainable competitive advantage
(Mathieson, Peacock, & Chin, 2001).
Tangible resources: on the other hand, these resources are physical assets, which includes machinery, buildings, land, physical capital and equipment. Such assets can be found and bought in the market easily, which makes them less competitive to the companies because their competitors and rivals can also acquire them (Mathieson, Peacock, & Chin, 2001). The company’s information systems are examples of such kinds of resources, because they are made up of a network of technological structures, which include computers, servers, and database machines, among others.
An important element of a company’s resources, in order to be competitive, they need to be both heterogeneous and immobile (Thong, 2001). Heterogeneous means that the information systems’ capabilities, skills and other IT resources that a company has should strongly differ from other companies (Ranilla et al., 2014). After all, if all companies had the same mix and amount of information technology systems, they would not need to use different strategies to compete and would simply follow each other without the achievement of any competitive advantage
(Peteraf, 1993). For instance, speaking of High Performance Computers, the author Ranilla et al (2014) explains that “Today’s HPC platforms are based on clusters of multicore nodes, which include accelerator devices such as GPUs or Xeon Phi coprocessors. Hence, the heterogeneity characterizes modern HPC platforms” (p.1). Therefore, the two critical assumptions of RBV are that resources must also be heterogeneous and immobile. It is important in today’s real-world markets, which are far from being perfectly competitive, that companies apply different IT strategies in order to sustain a competitive relationship with other companies in the market (Thong, 2001).
Examples of such competition comes from the companies of Apple and Samsung Electronics, which operate in the same industry, exposed to similar external resources but achieve different organizational results, because they use different bundles of IT resources within their disposal and they apply different strategies in utilizing their IT resources too. Samsung competes with Apple in the field of smartphones and tablet products, yet Apple uses its IT systems and resources in a way that allows it to sell its products at very high prices, therefore generating higher profits. On the other hand, Samsung follows a different strategy utilizing its information systems and resources in a different way, producing a longer and more varied line of products at relatively cheaper prices.
A second consideration towards resources is that they should be immobile and innovative. In other words, the company’s resources and information systems should be distinctive enough and immobile in a way that does not make it easy to move them from company to another company, at least on the short run. This immobility would allow companies to protect themselves from having their resources being replicated by rivals, who would then apply the same strategies. This is why it is also important for companies to capitalize on both their information systems and also link
them to their brand equity, knowledge, processes and intellectual property, which are all immobile types of resources. As Barney (1991) explains: “Effectively, this translates into valuable resources that are neither perfectly imitable nor substitutable without great effort” (p117). Other than the previously mentioned criteria for a firm’s resources, there are additional key criteria which would make an organization’s information system resources highly competitive on the long run, according to Makadok (2001). Primarily, a firm can sustain its competitive advantage above average returns if it ensures the following key steps are followed: First, it should identify its potential key IT resources. Second, it should assess whether these information system resources fulfill the following key criteria:
- Valuable: the information system resource should allow the firm to utilize a strategy that creates value, by either performing better than its competitors or reducing the weaknesses which it has (Makadok, 2001)
- Scarce: To be of tangible value, information system resource should be scarce by definition, which allows it to hold long-term value and increased future returns for the company as a whole.
- In-imitable: For a resource in the company’s information system to be competitive, it should also be under the full control of the firm, otherwise it would neither be scarce nor would it sustain its value. Such an advantage can be sustainable only if the firm’s information system asset cannot be perfectly duplicated, and so it should be in-imitable. (Peteraf, 1993, p183). Here, Mathieson, Peacock & Chin (2001) explain that a critical factor of inimitability of resources is the causal ambiguity, which happens if the firm’s competitive advantage has an unknown source or its information system’s high
quality is the work of its own operators, staff and employees, or comes from a custom-made and confidential source (Peteraf, 1993). If the resource in question is knowledge-based or socially.
- Non-substitutable: Finally, an important element that makes a company’s information system resources highly competitive and protected is that it cannot be substituted. This is because even if the information system is value-creating, scarce, inimitable, it will lose its competitiveness if competitors can easily provide an alternative to it. (Makadok, 2001). They can counter the value-creating information system of the firm by providing one that delivers a similar value yet in a lower price which can draw market share and consumers away from the firm (Amit & Schoemaker, 1993).
C. Diffusion of Innovation (DOI) Theory: Its stages, implications and limitations in the organization:
Developed by E.M Rogers back in 1962, the “Diffusion of Innovation (DOI) Theory” is considered one of social sciences’ oldest theories. Its main objective is to explain how a product or an idea gains momentum over time and spreads (or diffuses) through a specific target market or social system. The diffusion’s end result is that the people in that market where the product or idea developed adopt the behavior, idea or product
Adoption means that a person does something differently than what they had previously (in our case, acquiring and performing a new behavior as a result of a newly installed Information System within the organization). Yet, Rogers (1962) explains that the key to a successful adoption is that the employee should first perceive the new behavior, idea, product or system as something innovative, new and useful. It is through such a
perception that the diffusion of a new information system within the company is made more possible and more effective, especially that new changes are often faced with resistance at first.
However, it should be noted that, according to Rogers (1962), the adoption of a new service, procedure, or idea in the organization, does not happen simultaneously. Rather, it is a process whereby some individuals and staff members are more apt to adopt the new system than others in the same organization. The reason for that, according to researchers, is that different employee sand staff members have different characteristics, and so when promoting a new information system to a target group in the organization, it is important to properly understand the features of the target group which would either assist or hamper the adoption process of the new system. According to Rogers (1962), there are “five established adopter categories” to pay attention to. Accordingly, there are various strategies to be followed in an attempt to appeal to different adopter categories:
1. Innovators: These are the type of employees who want to be the first to adopt the new innovation or information system. This is because they have a character that is adventurous and always on the look out for new ideas. Such individuals are always ready to take risks and be the first to develop new concepts and ideas. To appeal to this target group, very little effort is needed.
2. Early Adopters: These types of employees are like opinion leaders. They enjoy certain leadership roles, whether it was formal leadership (from their position) or informal leadership (influence without a formal position). As such, these kinds of individuals embrace opportunities for change and innovation because they are already aware of the necessity to change and are generally
comfortable accepting and getting used to new ideas. To appeal to such a target group, we need how-to-manuals, guides and information sheets on how to properly implement the new Information System tactics and strategies. Information is not needed to convince them to embrace change, but rather practical guidance (MacVaugh, 2000).
3. Early Majority: These types of employees are not always leaders, yet they are able to adopt new ideas earlier than the average employee. That being said, the early majority individuals usually at when they see evidence that the innovation properly works before they decide to adopt it. In order to appeal to such a target group, the organization should follow the strategy of showing success stories and relevant evidence that prove the effectiveness of the innovation (MacVaugh, 2000)
4. Late Majority: These types of employees are usually skeptical of any change, and will only accept a new innovation in the organization’s Information Systems after it has been accepted and applied by the majority. To appeal to this target group, it is important to provide detailed information on how other employees adopted and applied the new system successfully and benefited from it (MacVaugh, 2000).
5. Laggards: Finally, the laggards, according to this theory, are employees that are very conservative and feel comfortable only following the traditional and conventional procedures. Therefore, such a group of employees are the most challenging to bring on
board when wanting to spread a new innovative information system in the organization. To appeal to this target group, an organization or a team leader should use strategies such as fear appeals, strong statistics and even pressure from other team leaders from other adopter groups (MacVaugh, 2000)
The figure above summarizes Roger’s diffusion theory visualizing the position of every type of adopter.
There are stages through which an employee in an organization would go through to adopt a new innovation in the company’s information system, before which diffusion is fully accomplished. The first stage is the awareness by the employee that there is a need for an innovation, followed by his or her decision to either accept or reject the adoption of the innovation. In case there was acceptance, the third stage would be the initial use of the innovation, which is a stage in which the innovation is tested. If successful, the fourth stage of ‘continued use’ takes place. Generally speaking, no stage or process of adopting innovation in the organization’s information system continues without influence. According to the theory, there are five key factors which affect the adoption process of an innovation, each of them affects in a different extent:
1. The extent to which the new system is perceived by the employees as a better system than then one that it replaces. This is known as its Relative Advantage
2. A second key factor is the extent to which the new innovative system is in line with the company’s experiences, values and needs of its employees. This is known as Compatibility.
3. Thirdly, would the new system be difficult to use or understand? This will affect whether it will be adopted or not, and whether it will spread or not. It is known as Complexity.
4. Fourth of all, can the new system be experimented or tested before committing to adopt it? This is known as its Triability.
5. Finally, employees would care to know the extent to which the innovation provides results that are tangible and measureable. This is known as its Observability.
Limitations of the “Diffusion of Innovation” Theory
It should be noted that the ‘Diffusion of Innovation’ theory has been successfully used in many sectors, including agriculture, communication, criminal justice, public health, marketing and social work (MacVaugh, 2000). Yet, despite its advantages and ease of application in some stages, the diffusion of innovation theory does have limitations that need to be considered. The main limitations include:
- The adopter categories are not applicable to every single organization in every sector. Certain sectors, such as the public health sector, face this challenge. This is because according to the Boston University School of Public Health (2013) most of the evidence of this theory, mainly the adopter categories, neither originated in public health nor was it developed to apply to the acceptance of health innovations or new behaviors. Also, the theory does not include a participatory procedure of adopting a public health-related information system.
- The theory works more effectively with adopting new behaviors and new systems, more than preventing or stopping old or bad behaviors and systems.
- The theory does not consider the social support or resources available for the individual when adopting a new system or innovation in the organization.
How can a service provider leverage their resources to ensure that a potential HPC user finds it useful and is able to use it? And when they do, how will the user react to the offered service?
III. Case Studies
In this section, we will elaborate more on the HPC technology and the provider of HPC infrastructure in the Netherlands, then present a few case studies that have been perused to shed the light on how potential SMEs perceive the usefulness, ease of use, incentives and barriers of the adoption of HPC technology. We will examine success stories and failed attempts and later on, we will take a closer look at the marketing efforts that HPC providers are utilizing to promote the technology, whether it was through collaborative projects, awareness programs or by prototyping.
At the outset, what do we know about HPC?
High Performance Computing most generally refers to the practice of aggregating computing power in a way that delivers much higher performance than one could get out of a typical desktop computer or workstation in order to solve large problems in science, engineering, or business. (What is high performance computing?, 2014)
The term HPC is occasionally used as a synonym for supercomputing, although technically a supercomputer is a system that performs at or near the currently highest operational rate for computers (Ranilla, Garzon & Vigo-Anguiar, 2014).
“A helpful way to help understand what high performance computers are is to think about the what’s in them. You have all of the elements you’d find on your desktop — processors, memory, disk, and operating system — just more of them. High performance computers of interest to small and medium-sized businesses today are really clusters of computers. Each individual computer in a commonly configured small cluster has between one and four processors, and today’s processors typically have
from two to four cores. HPC people often refer to the individual computers in a cluster as nodes (What is high performance computing?, 2014). A cluster of interest to a small business could have as few as four nodes, or 16 cores. A common cluster size in many businesses is between 16 and 64 nodes, or from 64 to 256 cores.” (What is high performance computing?, 2014)
The figure above shows how Big Data available on hand are processed by HPC and delivered as information to the end user. Ref: [3]
HPC was introduced in the 1960s with the release of the “CDC 6600 supercomputer” followed by the famous Cray 1 in 1976, which became the most successful supercomputer in history (What is high performance computing?, 2014). HPC performance is measured in FLOPS (Floating-point Operations Per Second). The first supercomputers could sustain a few hundred kiloFLOPS (103 FLOPS); today, the national supercomputer of
China Tianhe-2, was ranked the world's fastest with a record of 33.86 petaFLOPS (1015 FLOPS). Another way to put things into perspective, in 1961, the approximate cost per GFLOPS (gigaFLOPS) was U.S$8.3 trillion. Today, the cost is about $0.08.
Traditionally, HPC has been mostly employed in the highly sophisticated applications in the world. It “has become an essential tool in both scientific and industrial communities because it provides an excellent environment for solving a wide range of problems in modern science and engineering.” (What is high performance computing?, 2014) Applications include biosciences, quantum physics, weather forecasting, oil and gas industry modeling, physical simulation, electronic design and automation, airplane & spacecraft aerodynamics, universities (basic and applied research), government labs, digital content creation, and more... (What is high performance computing?, 2014)
According to Earl Joseph, the head of Technical Computing practice at IDC, Supercomputer revenues actually accelerated during the global economic downturn, driven by the growing recognition of the crucial role these systems play in economic competitiveness as well as scientific progress (Ranilla, Garzon & Vigo-Anguiar, 2014).
This figure shows the adoption of HPC in time by different sectors (Governmental, Academia and corporate) beginning in the mid 19th
century until today.
Why is it important that we know the barriers and incentives for SMEs to adopt HPC? First, lets discuss the relevancy of this technology in the business world, and what will be the impact of its applications on the Small and Medium-sized Enterprises, particularly in Europe.
To obtain more accurate information about HPC, we have conducted an interview with Anwar Osseyran, Managing Director of SURFsara. Below are his answers to the main questions asked about his company, its mission and competitive edge.
In explaining why does he think his company has a competitive advantage, among other service providers, in motivating SMEs into adopting HPC instead of them relying on their own in-house resources? Osseyran (2015) explained that SURFsara’s mission is to support researchers in academia and industry in The Netherlands by developing and deploying state-of-the-arts HPC infrastructure for R&D. The company has therefore developed a wide range of HPC architectures and related expertise in the past forty three years of its existence. The budget needed for maintaining the Dutch HPC ecosystem is mainly provided by the Dutch government from education and research funds but also from the ministry of economic affairs as the competitive benefits of e-infrastructure in general and HPC in particular are becoming more apparent. Other funding sources are provided by the EU through European collaborative projects, research contracts with Dutch industry and technology transfer funds as there is a growing tendency to deploy HPC to increase market penetration
academia and industry, mission to support both, elaborated expertise in HPC technology and application domains, financial support from the Dutch governmental agencies and installed base of projects and collaborations make SURFsara the ideal partner to Dutch companies when they deploy HPC in their R&D.
In his experience, Osseyran (2015) considers that the ultimate decision to adopt a technology such as HPC depends on the company’s organizational resources and/or corporate strategy. This is because on the industrial side, large Dutch companies fund their own equipment, albeit, in general, large cluster-type systems, that is, good for high throughput computing but less suited for large-capability jobs on many cores (Osseyran, 2015). These large companies (such as Shell, Akzo, Unilever, and Philips) usually have their own research facilities in the Netherlands, but companies in the banking and insurance sectors are also increasingly using HPC for portfolio simulations and mathematical insurance modeling. The research collaborations between large companies and Dutch academic research groups are generally well established, leading to indirect access of large industry to the national HPC facilities (Osseyran, 2015).
The missing middle between individual HPC users and large corporate HPC applications is very clear in The Netherlands: SMEs rarely have facilities larger than workstation or multicore powerful PCs. There is a lot to be gained here, but this has to be organized. A significant amount of effort will need to be put in both offering hardware and software services and in modeling, simulation, and application development (Osseyran, 2015). So to recapitulate, large multinationals know their way to use HPC and they do it frequently through their collaborations with academia. SMEs need to be more stimulated and supported as they mostly have a short term scope and less financial reserves to focus on mid-long term R&D (Osseyran, 2015).
Osseyran (2015) also believes that the HPC Service Providers play a marketing role in diffusing its technologies, especially in new market segments. This is because HPC is becoming the differentiator between good and excellent products and services and as SMEs need to be stimulated, and so HPC service providers must indeed promote and diffuse HPC. Future societal challenges in The Netherlands require a strong commitment to the deployment of HPC for scenario development and analysis. The physical presence of HPC infrastructure and facilities is required for solving major social problems, promoting key areas in computational science, and strengthening economic structures through innovation and creating jobs (Osseyran, 2015). The Dutch government recognizes that supercomputer infrastructure and services have positive societal and economic impact and lead to competitive advantages in various industrial fields as well as help train less skilled personnel. HPC service providers have to convey this message and to make HPC available to all.
In responding to the question if there are alternatives to the traditional way of using HPC (maybe like HPC Cloud, etc…) and their effectiveness, Osseyran explains that upfront and ongoing costs of HPC form serious barrier and reduce also the ability to scale once adopted. As HPC needs fluctuate, it’s difficult to adequately size the needed infrastructure and related resources. This means resources will either be idle at times or insufficient. On the other hand, HPC evolves quickly and this means the need to continuously follow the technology and update the hardware otherwise a performance gap may get result with the competitors. Next to hardware investments, HPC demands specialized power installations that consume a lot of electrical power and require specialized datacenters and technical staff (Osseyran, 2015). Offering HPC services through the clouds become indeed attractive to many companies. But we make a distinction between Capability and capacity HPC services. Unlike capability
provided through the clouds and does not require a lot of personalized services from HPC centers. Osseyran also added that HPC cloud services are in contrast not always ideal for engineering and scientific applications requiring capability computing. HPC cloud providers, such as Amazon AWS, do not offer tightly-coupled HPC networks that engineering and scientific applications require. User support is also limited and users have often to manage their own virtual environment (installing all applications, testing, trouble shooting, etc.) or are required to learn new interface and workflow processes which are time consuming. To summarize, in Osseyran’s view, HPC cloud services offer in many cases to SMEs a much better alternative to own HPC infrastructures when it comes to production-like services where HPC is involved. For state-of-the-arts R&D developing prototypes and getting high-tech competitive products faster on the market, HPC capability centers may be the better alternative.
Finally, Osseyran concludes that what indeed most counts for SMEs is that HPC services are easy to access and adequately respond to their needs. For engineering companies for instance, HPC services should offer computer engineering tools for modeling and simulation such as Computational Fluid Dynamics (CFD), Finite Elements Analysis (FEA), stress analysis, thermodynamics, structural mechanics and many others at the highest level of performance (Osseyran, 2015). Application areas include weather forecasts and environmental simulations, energy exploration, life sciences and R&D all demanding vast numbers of data points for accuracy. Nevertheless, having adequate HPC resources and easy access to those are two necessary but not sufficient conditions. A total eco-system is needed especially to ensure that continuity of services is guaranteed and that academia and industry continue to collaborate in advancing science and engineering and preparing the needed human resources for the future (Osseyran, 2015).
Why does HPC matter?
Besides from HPC’s role in predicting dangerous weather patterns, designing safer, more fuel-efficient vehicles, and solving the secrets of the universe, HPC is a tool that can boost business performance and efficiency.
a. Time and Efficiency
HPC is nowadays evolving quite fast attempting to meet the demands for faster computing and higher processing speed, in most sectors including science. HPC has proved to reduce time-to-market for many businesses around the globe, and consequently improved their efficiency (Welsh & White, 2014). Reducing the execution time of some applications and faster calculations not only serves the direct user (i.e. engineers) but also the company; and saving the engineers time, means saving the company money, which means higher Return on Investment.
b. Expand the boundaries
HPC’s role isn’t limited to predicting dangerous weather patterns, designing safer, more fuel-efficient vehicles, and solving the secrets of the universe, but also to find and exploit weaknesses in financial markets or analyze customer behavior (What is high performance computing?, 2014). It enables higher accessibility for employees and multi-user access to different graphics-heavy and business-critical applications (What is high performance computing?, 2014). Software scalability and crossing the limitations in codes scalability has became one of HPC’s most technical
c. Sustainability
The World Wide Fund For Nature (WWF) and the American Council for an Energy Efficiency Economy (ACEEE) quoted that “For every extra kilowatt-hour of electricity that has been demanded by ICT, the U.S economy increased its overall energy savings by a factor of about 10.”
Even though studies have shown that ICT is responsible for 2% of the carbon emission in the world (Prout, 2015), HPC providers are currently implementing HPC cloud computing which is a virtual infrastructure having the same advantages in addition to “green IT” (Prout, 2015).
We showed in the paragraphs above how HPC is not only working on the frontier of science, but is also serving the industrial world by reducing the time-to-market which represents a crucial element in the profitability of a business. Sometimes, making use of such a complex technology could be a real challenge, and although the usage of HPC has increased drastically in the past decade, some businesses and especially small and medium-sized ones lack the guidance and expertise in employing HPC to improve their performance or enhance the quality of their products and services (Welsh & White, 2014).
The Service Provider: SURFsara
Typically, in order to use HPC, an infrastructure of systems with massive number of processors, grids, network, data centers and cooling systems is required. The computing systems generally are developed by hardware manufacturers (such as IBM, Bull, Cray…). This infrastructure is usually owned and maintained by a company that offers the HPC technology as a
service for the industrial and scientific communities to use it, by designing tailor-made software systems according to the user’s needs. This company is hereby referred to as “Service Provider”.
In the Netherlands, SURFsara, is a Service Provider that offers an integrated ICT (Interactive Communication Tools) services, infrastructure, in the areas of data storage, computing, cloud, networking, visualization, e-science and most importantly, multidisciplinary expertise and support in ICT technology and applications to the clients and end users (Surfsara, 2014)
SURFsara owns several HPC systems. For example Cartesius, a clustered symmetric multiprocessing system built by Bull is the Dutch national supercomputer. It is currently in the TOP500 supercomputers in the world now at position 421, and ensures maximum usage by the Dutch scientific community (Ranilla, Garzon & Vigo-Anguiar, 2014). The Lisa Cluster on the other hand is another system of large computing capacities, now being used by the University of Amsterdam, the VU University Amsterdam and the NWO; in addition to several a number of other HPC systems that are dedicated to the support of research in the Netherlands.
SURFsara faces its own challenges. We previously mentioned that SURFsara is one of the very few, if not the only one, to provide HPC packaged with usage consultancy to support its client in figuring out the most effective way to make use of HPC to improve their products and services.
In my case study, when I asked companies whether or not they think that offering expert’s support by the Service Provider is necessary, 90% agreed that it is.
SURFsara has a highly competent team, experienced in employing HPC into new profitable business models. However, sometimes difficulties arise when customers try to communicate their business vision and what result they expect out of the service (Osseyran, 2015). Synchronizing the customer’s desire with the possible outcomes of HPC can be quite a challenge sometimes (Osseyran, 2015).
In one of the studies made by the Fortissimo Project, it is said that "Namely, when the IT managers of this operation described it, they were focus on the mind shift, not on the results (Sloan, 2015). It was explicitly said, that with a normal HPC center or cluster, you have to think about translating the problem on the resources that are available (Sloan, 2015). In other words, it’s more about how to use the infrastructure as best as possible and how to adapt the model of computation to this infrastructure. On the other hand, using Amazon’s AWS services, they were focused only on the problem, and the speed of gaining results, as there were no limits with regard to the infrastructure. And this shift from thinking about the infrastructure constraints to thinking about just obtaining results is the most important indicator of the paradigm shift that is also represented by the Google trends keywords. People stopped thinking about infrastructure and started thinking about its usage." (Sloan, 2015).
We will further look at how SURFsara is leveraging its resources in order to diffuse HPC onto new markets and hence maximize its usage among SMEs for example.
Methodology
In order to obtain accurate data and to understand better the key drivers and barriers to the adoption of HPC by SMEs, we decided to utilize both qualitative and quantitative methodologies in our research.
We conducted two interviews: one with a small company that adopted HPC and another with a medium-size company that briefly used HPC then ceased. In another dimension, we surveyed a larger number of SMEs that have adopted HPC and we asked them about their experience in using HPC. The SMEs are participants in the Fortissimo project which we will talk about later on, and the questionnaire, which is designed to suit the context of the study, serves as a quantitative approach to evaluate the constructs of the TAM model: Perceived Usefulness (PU), Perceived Ease of Use (PE), Attitude (AT) and Behavioral Intention (BI).
Diffusion of Innovation
As part of its strategy to diffuse the HPC technology into new market segments, and among a new group of users – more precisely Small and Medium-sized Enterprises, SURFsara who is currently an active partner in PRACE, has launched the HPC Exchange program and is working on the Fortissimo project. Below are details about the 3 main marketing arms of SURFsara, which aim to recognize (and nevertheless cross) the barriers and proclaim the benefits of High Performance Computing.
SHAPE by PRACE
SURFsara is the Dutch member of the Partnership of Advanced Computing in Europe (PRACE) since its establishment in 2010, along with 24 other members, each from a different European country.
An international non-profit association, PRACE is also a research institution which provides top-level HPC (high performance computing) service for both researchers and scientists from the industry and academia in Europe, and allows for engineering research and scientific discovery of high impact through all disciplines to improve the European
One of the new pan-European programs supported by PACE is SHAPE (SME Adoption Program in Europe). This program aims to provide the European SMEs with the needed expertise for them to benefit from innovation possibilities provided by HPC, in addition to increasing their awareness, and in this way, they would be able to increase their level of competitiveness.
SURFsara and its fellow HPC service providers in Europe contribute to this program by selecting applications for SMEs to a “Pilot Call”. The SHAPE Pilot is a free trial phase of the program and it aims to work with the selected SMEs in order to help them adopt HPC. According to the 2013 annual report of PRACE, 10 European SMEs were selected for the SHAPE pilot (SHAPE Program, 2014).
In PRACE’s website, it is mentioned that “The Programme will help European SMEs overcome barriers to using HPC, such as cost of operation, lack of knowledge and lack of resources. It will facilitate the process of defining a workable solution based on HPC and defining an appropriate business model.” (SHAPE Program, 2014)
HPC Exchange
As part of its efforts to converge HPC with small and medium size businesses, SURFsara has implemented a program called “HPC Exchange” that serves to raise awareness of the opportunities that HPC offers among SME entrepreneurs, contribute to accelerate the innovation of products and services and contribute to cost savings by lowering the cost of ownership through HPC cloud services (Prout, 2015).
The policy of Dutch government is now more focused on, strengthening the position as a digital gateway to Europe, and on encouraging the use
of advanced ICT so that companies can deliver greater value to their customers and to build a temporary competitive advantage (Prout, 2015).
In view of this, the Ministry of Economic Affairs developed several national initiatives, including the Breakthrough project SMEs innovating. SURF is one of the partners in this project.
The ambition in this regard, among other things, is to provide targeted incentives for products and services innovations in SMEs, through the efforts of the High Performance Computing facilities at SURFsara.
The Exchange offers training, specialized expertise, software solutions, access to knowledge, best practices, shared solutions and creates internal communication about this (SHAPE Program, 2014).
The key success factor for the application of HPC for innovative activities is not so much hardware, but especially software and services. An investment in infrastructure should be accompanied by investment in software and expertise
The HPC Exchange is the place where HPC infrastructure, software and services are offered to integrate knowledge institutions and companies. HPC Exchange acts as an enabler to:
• Business innovation
• Cross-sectorial cooperation between enterprises
• Connection between scientific knowledge and innovation in companies
The HPC exchange can play a meaningful role in the value networks that arise from the cross-over collaboration, by offering processing power of knowledge institutions and private organizations.
The quest to understand the barriers of HPC usage by SMEs is still in play, till this day. Service providers like SURFsara are yet to unveil obstacles and figure out ways to overcome them. Deriving from this mission, SURFsara has set up a strategic plan to facilitate the acquittal of HPC technology by SMEs that are able, possessing this high-tech tool, to maneuver a competitive difference in the market.
This plan consists of a 3-phases system as following:
1- Pilot
Which is a free trial (prototype) phase that enables SMEs to gain insights of the capabilities offered by innovative ICT solutions such as HPC and the steps required to achieve a sustainable business solution.
2- Proof of Concept
Which is a paid research assignment, where the client has a clear vision of what he wants. It is basically the development and optimization of a business application.
3- Implementation
The final stage, where an application is converted into a marketable solution using knowledge and technology acquired during the course of the project. At this point, HPC can become a permanent feature of business operations.
The advantages of a HPC Exchange for the different participants can be summarized as follows:
1. Benefits for companies:
• Easy access to HPC facilities with no upfront investment • Cheap or free pilot to try to use HPC
• Training in use HPC
2. User support:
• Availability existing software solutions
• Access to commercial partners for specific questions / problems • Access to specialized expertise
• Access to relevant scientific knowledge
• Access to shared knowledge and HPC solutions
• Contributions to or available quickly gain of technological innovations
• Technology watch
3. Benefits for scientists:
• Access to collaborative projects with companies • Better HPC facilities by sharing with companies
4. Advantages for the government:
• Better utilization of resources for the competitiveness of businesses • More cooperation aimed at innovation
• Better HPC facilities for science by sharing with companies • Coordinated deployment of national resources
• Encouraging public - private partnership
• Building and maintaining strategic expertise in computational science
• A strong leverage effect of investments in HPC infrastructure.
Fortissimo project
According to Sloan (2015),the Fortissimo Project is a project that is of collaborative type, and will allow the Small-to-Medium enterprises in Europe to be more competitive on the global level through their use of simulation services running on HPC cloud infrastructure (Sloan, 2015). SURFsara is an important partner in that project, which is coordinated by the University of Edinburgh. The project also includes a number of partners, including Application Developers, Manufacturing Companies, Domain Experts, HPC Cloud Service Provider, and IT Solution Providers from as much as 14 countries. The European Commission funds the Fortissimo project, within the 7th Framework Programme (Sloan, 2015)
Fortissimo’s main goal is to overcome this impasse. Doing so should happen by the provision of simulation services and tools running on a cloud infrastructure. Sloan (2015) explains that what would greatly ease the access for SME’s towards advanced simulation is the “one-stop-shop” concept, which will make expertise, applications, hardware, visualization and application tools easily available and affordable on a pay-per-use basis. By doing so, Fortissimo would be able to prove a sustainable and viable commercial ecosystem (Sloan, 2015).
On July 8th 2015, A Fortissimo Forum took place in Amsterdam joining
SURFsara (being one of the main sponsors of the project), along with other European HPC Service Providers, ISVs and more than 20 SMEs.
A survey has been distributed to most of the SMEs that participated in the project, in attempt to gather data and draw conclusions on how they perceive the usage of HPC (Sloan, 2015). The results of the survey in “Exhibit A” reveal how SMEs are susceptible and receptive of the technology, and that will be elaborated more in the Analysis section.
The User: Small and Medium-sized Enterprises
Now, we will look at two different companies that each has adopted HPC, one that successfully implemented the service and is still using it, and another one that used HPC for a short period of time and decided to drop out. I conducted the following two interviews with directors from each SME to look at adopting HPC from a user point of view.
MicroLife Solutions: A small enterprise with huge potential
Founded in 2011, MicroLife Solutions is biotechnology startup company that aims to explore the natural diversity of bioactive substances and enzymes for applicability.
Although MicroLife Solutions is considered a small enterprise, it was keen to undergo the Pilot experience that SURFsara had to offer anticipating that it will be an added value to their biotech research activities (Irza, 2012)
In the beginning of 2015, MicroLife Solutions acquired a limited number of "CPU hours” as part of the Pilot project (Irza, 2012). The company had selected the VM-HPC Cloud (Virtual Machine on HPC Cloud) service.
According to the MIT Lincoln Laboratory in Massachusetts - USA, “those who have adopted virtual deployment environments have capitalized on a variety of advantages including server consolidation, service migration, and higher service reliability. But they have also ended up with some challenges including a sacrifice in performance and more complex system management” (Irza, 2012)
According to Prout (2015), MicroLife is using VM on HPC cloud, mainly because they don’t want to wait for queuing, and HPC Cloud solves this problem, despite the numerous challenges associated with its usage. However, Prout (2015) also indicates that some companies still prefer its advantages that it has to its alternative. The technical terminology of the usage of HPC Cloud is Infrastructure as a Service (IaaS), and this service requires the end client to have enough IT expertise to use it properly, which is why some companies prefer to expand their IT department hire experts than having to face the queuing barriers of the traditional way to use HPC, the Platform as a Service (PaaS). (Prout, 2015)
I conducted an interview with Thierry K.S. Janssens, Ph.D., Unit Director in MicroLife Solutions, and when asked about the incentives, how the service improved their products and the barriers behind the adoption of HPC, Janssens said:
SURFsara offers a scalable and reliable computing service. Academic partners of us often use it. We had the possibility to disclose some of the sequence information of samples of interest that allows us to develop new products (Janssens, 2015). However, the myriad of bioinformatics techniques, the basic knowledge of informatics and the underestimation of computing needs and storage are hurdles that have to be taken (Janssens, 2015).
And on its perceived ease of use, Janssens commented: The consultant helped us setting up the VM so that we just had to install the software we needed (Janssens, 2015). The use of remote desktop allowed easy access from within our domain.
BaseClear: Where the difficulties prevailed
BaseClear is an independent service laboratory that provides DNA sequencing and analysis services to academia and industry. With a little
more than 50 employees, BaseClear is considered a medium-sized enterprise.
Currently BaseClear isn’t using HPC, but in 2014 they have collaborated with SURFsara through a joint ICT Breakthrough project after being motivated by the need to gain more knowledge about HPC computing given that this would be a key element for sustainable analysis of (ever growing) large scale DNA/genomics analysis, as Walter Pirovano – The Director of Bioinformatics Dept. at Base Clear said.
When asked if the usage of HPC by BaseClear improved its performance, Pirovano said that it did not really improve BaseClear’s performance however they had a better insight into the advantages and limitations (Janssens, 2015).
He added, “In practice it turned out to be too difficult for our Bioinformatics team to setup a workable environment. So in the end we did not use SURFsara’s HPC computing service. Even though we believe it could be useful, however after our first trial, we concluded that we need to enlarge our internal computing capacity to remain independent, not to mention that from a cost perspective, this is more convenient.” (Janssens, 2015).
IV. Findings & Analysis
SURFsara has been working on promoting High-Performance Computing to small and medium-sized enterprises for years, and it has already done a few market analysis studies, along with its partners in the field, in attempt to figure out where HPC stands in the SME market, and how potential adopters look at this technology from different points of view.
By going through the archival data of SURFsara’s most recent market analysis studies, I came across several insightful results.
For instance, Table 1 hereunder shows the financial ROI results of large, medium sized and small organizations. By looking at the table, we notice that the medium-sized company is the hero of the ROI. It brings tremendous revenue and profit per HPC Dollar invested.
When it comes to innovation, large enterprises are actually the ones investing the most in HPC, almost 6x more than SMEs.
Actual figures (In million USD) are shown in Table 2 here:
Table 2. HPC investment per innovation by Organization size
A questionnaire about marketplace benefits has been distributed among SMEs and we have asked questions to which we received the following results:
1. How do you/would you choose among different infrastructure providers?
2. What are the barriers of adopting outsourced resources (HPC)?
This fundamental question came with varied answers. SURFsara works on overcoming the “Lack of expertise” and “Lack of information” barriers and shows that even though it costs more to outsource high-performance computing, the return is worth the investment.
3. In what way would such marketplace bring return on investment?
In the Chart 3 above, the answers “Reduce costs of IT support” and “Reduce time to market” are correlated with the problems of usage of in-house infrastructure.
4. We inquired about the satisfaction of the available in-house resources. The results are as follows:
This chart shows that the majority of companies are not satisfied with their in-house computing resources, which pushes them to seek
5. Why wouldn’t companies rely on their computing capacities?
Disadvantages of using in-house resources
6. What are the advantages of using in-house resources?
7. …And regarding the barriers for adoption of outsourced HPC infrastructure:
Allowing simple experimentation and access to experts is therefore crucial for the decision making of the customer upon outsourcing HPC infrastructure usage.
By analyzing the data obtained from the tables and charts of this market analysis study, we can see the points of difference and pros/cons between in-house resources and outsourcing computational power (generally HPC), and thus extracting the following findings:
1. Companies that use in-house computing resources instead of outsourcing them do it for financial reasons most essential, and also because keeping it in-house allow them to have complete control of the overall process.
2. Since in-house resources only cover basic examples, either due to lack of computational power or lack of experts, companies are generally unsatisfied with these capacities and seek to utilize HPC infrastructure from its providers.
3. HPC providers are selected based on recommendations from partners and friends, to a lesser extent on systematic research.
4. Complete solutions (HPC, experts, license, etc.) are preferred over having to deal with them separately,
5. The data should be secured and HPC in the cloud should be affordable.
6. Marketplace is seen mostly to reduce the time to market the product and to reduce costs of IT support.
Barriers to adopting High-Performance computing by Small and Medium-sized Enterprises may be diminished by offering experience through experiments and free or low-cost trial runs and expertise through support, webinars, trainings, innovation communities contributions, etc. Even when it comes to the cost, prototyping provides the adopter with a better idea on whether or not the implementation of HPC would induce relatively higher returns on investment.
As previously mentioned, SURFsara’s HPC Exchange Program has opened the door to several SMEs, that have limited resources and expertise, to make us of HPC in order to improve their products and services. MicroLife for example, was able to develop its scientific research capabilities using HPC after being able to experience the technology with the Pilot phase of the HPC Exchange Program. With the experts’ support of SURFsara, MicroLife was able to overcome difficulties related to expertise and
experience, which according to our findings are two of the most popular barriers for outsourcing high-performance computing.
In pursuance of better understanding the perceived usefulness and ease of use of HPC by its users, I designed a survey that calls upon the constructs of the Technology Acceptance Model and asked 25 European SMEs to take it, on anonymous basis.
After analyzing the results of the survey shown in Appendix “A”, we can draw the following conclusions:
1. Most SMEs think HPC is not easy to use, independently, without experts’ support. Experts’ support is necessary.
2. Most SMEs strongly believe that HPC improve their company’s business performance, enhances the quality of their products/services, and gives the company a competitive advantage by reducing time-to-market (Welsh & White, 2014).
3. SME’s attitude towards the usage of HPC is positive, and most SMEs think adopting HPC is a wise idea.
4. Answers varied when asked about their behavioral intention to use. For example some agreed and some didn’t when asked if they would use HPC exclusively in complex matters. But most users affirmed that they would use HPC all the time.
V. Conclusion
In conclusion, in today’s increasingly competitive and demanding information age SMEs are inevitably heading towards results that are quicker, more reliable and more accurate through the use of various computer technology. The results of the paper have shown that High-Performance Computing (HPC) is one of the computing approaches which indeed carries with it a number of advantages for business in terms of efficiency, quality of products and services and a long term competitive advantage, which has led to a positive standpoint by the SMEs towards the use of HPC. More specifically, benefits were seen in product differentiation, internal and external communication, research and development, innovation and consequently increased ROI, providing the company with a sustainable competitive advantage. This is why HPC is being used by enterprise of all sizes and types, including governmental agencies, invoking positive impact on their performance. Yet the use of HPC has also shown to be challenging to for Small to Medium Enterprises without the presence of expert support, especially in complex matters.
The research also showed, based on the survey that was conducted, that the majority of SME managers choose their infrastructure providers based on recommendations from friends and partners, while others rely on research, which highlights the importance of previous experience and the role of stakeholders in the choice of technology. In adopting outsourced resources such as HPC the main barriers cited by managers are the cost, followed by the lack of information.
As for the potential return on investment, managers indicated that the use of HPC reduces the cost of IT support and the time to reach the market, which explains why up to 45% of the respondents are unsatisfied with the available in-house resources at their disposal, since they have a
