International Journal of Information Management 34 (2014) 221–225
Contents lists available at ScienceDirect
International Journal of Information Management
jou rna l h om epage: www.elsev ier .com/ locate / i j in fomgt
iewpoint
iscovering the potential of cloud computing in accelerating theearch for curing serious illnesses
abil Sultan ∗
niversity Campus Suffolk, School of Business, leadership and Enterprise, Ipswich, Suffolk IP4 1QJ, United Kingdom
r t i c l e i n f o
rticle history:vailable online 21 January 2014
eywords:loud computing
a b s t r a c t
Derided, when it emerged in 2007 as a fad, cloud computing has proved to be a viable business modelfor remotely delivering IT services through the Web (and other media) on a pay-as-you-go basis. Theflexibility of this emerging computing service has opened many possibilities for organizations. Drugcompanies and medical research centers are among those organizations that are likely to benefit from
igh performance computingNAife science
this new IT service model. As well as providing massive cost-savings, cloud computing could offer theseorganizations the opportunity to greatly enhance the efficiency of their operations. For cloud providers,this is a new field to expand their reach. The aim of this article is to explore this new development and thepotential of cloud computing in contributing to the advancement of research in life science and explainwhy this IT service model (despite many of its problems) could be game-changer for companies engagedin this business.
. Introduction
Cloud computing is an IT solution and a business model that usesdvances in ICT technologies such as virtualization and grid com-uting to remotely deliver (on demand) a range of ICT services (e.g.,usiness and development software, processing power, storage)hrough the Web and other media such as a network infrastruc-ure. There are mainly three types of services that are providedy cloud computing: Software as a Service (SaaS), Infrastructures a Service (IaaS) and Platform as a Service (PaaS). SaaS involveshe provision of software functionality (Google Apps is a good andopular example). IaaS provides a range of infrastructural servicese.g., processing power, storage, virtual servers). PaaS is a platformhat enables access to development software and hosting options toevelopers of Web applications. Cloud computing represents a newusiness servitization model that is different from those describedy servitization authors who saw a service as either a supplemento an existing physical product and/or a service that is based on
supplier using its skills and knowledge (i.e., its competences) torovide clients with a solution (see Baines, Lightfoot, Benedettini,
Kay, 2009; Vandermerwe & Rada, 1988; Vargo & Lusch, 2006,008). The cloud paradigm is different in the sense that a physical
roduct (e.g., software and hardware) is transformed into a service.
Cloud computing is underpinned by two main technologies: vir-ualization and grid computing. Virtualization can be described as
an approach for pooling and sharing technology resources to ensuregreater efficiency of resources utilization. For example, virtualiza-tion can be used to take a single physical asset (e.g., server, storagedevice or network) and make it operate as if it were many sep-arate, smaller assets. This process improves asset utilization andefficiency, and decreases costs by reducing the need for physicalassets. Moreover, virtualization can also be used to combine mul-tiple assets (e.g., storage devices and networks) and present themto servers and applications as if they were a single, larger asset;which simplifies server and application architecture and reducescosts (Sultan & Salim, 2010).
Grid computing is the technology that involves the use of soft-ware to combine the computational power of many different (andpossibly geographically dispersed) computers, connected in a grid(hence the name “grid computing”) in order to provide enhancedcomputer processing power. Grid computing also uses softwarethat can divide and farm out pieces of a program to as many asseveral thousand computers. Grid technology, therefore, can bethought of as the technology that enables the establishment ofnetwork-distributed parallel processing and distributed and large-scale cluster computing.
2. The cloud and high performance computing
It is interesting to note that cloud computing for high per-
formance computing (HPC) has not been on the priority list ofmany cloud providers. Until recently, HPC has not been a goodcandidate for cloud computing due to a number of factors suchas its requirement for tight integration between server nodes via
ow-latency interconnects and high-speed networking (Shainert al., 2010). For example, the performance overhead associatedith host virtualization, a prerequisite technology for migrating
ocal applications to the cloud, quickly erodes application scalabil-ty and efficiency in an HPC context which often involves sending
essages back and forth many times per second, a process thats likely to increase the possibility of latency (Niccolai, 2009).owever, new virtualization solutions that use KVM (kernel-basedirtual machine) and XEN hypervisors1 have managed to solvehe performance issue by reducing the virtualization managementverhead through enabling native performance capabilities fromhe virtual machines (VMs) and by allowing direct access from theMs to the network. High-speed networking is also an important
actor in HPC which requires fast communication between clustersf servers and storage (Shainer et al., 2010).
With the ability of more cloud vendors to provider faster speednd networking connections, cloud providers began to realize theusiness potential of providing reliable and cost-effective ser-ices for HPC-oriented organizations. Several application areas inhis category appear particularly suitable for cloud computing.mong those are drug discovery, personalized medicine, transla-
ional medicine and genomics (Rubenstein, 2010). Currently, therere few cloud providers that can declare themselves as having cloudolutions for these types of applications. Amazon is no doubt theominant player in this field. However, it is likely that more cloudroviders will be moving into this computing area as more peo-le from the scientific community begin to look at this computingervice for solutions to their problems.
. Life science research and development (R and D)
Advances in personalized medicine hold immense potential foruman health. The variation in human genetic make-up causes dif-
erent responses to drugs in many people. This is often why drugsail to treat many people as they are only suitable for a small pro-ortion of the entire human population (i.e., those with the “right”enome for a particular drug). The genome is the entire heredi-ary information of an organism (such as a human). Deciphering anndividual’s genome is a process known as DNA sequencing whichnvolves determining the order of the nucleotide bases2 – adenine,uanine, cytosine, and thymine – in a molecule of DNA. This is aomplex and time-consuming process; which explains why lesshan 20 human genomes have been sequenced so far in the entirecientific community (Reid, 2008). However, with the advent ofext-generation DNA sequencing machines there is potential forreating individual genetic “maps” that can be used to provideedicines tailored according to each person’s genome. However,
here is great potential for this process to be advanced through thenlimited processing power and storage of cloud computing.
Next-generation DNA sequencing machines create massivemounts of data and require massive amounts of processing power.ost of the IT infrastructures of pharmaceutical and biotechnology
ompanies are unable to provide this capability. For example, the
lient-server architectural approaches of these companies are oftenased on relational databases that are too expensive to develop,eploy and maintain, thus causing these companies to increas-
ngly seek commercial solutions to avoid unforeseen scaling risks
1 A hypervisor is a virtual machine monitor (VMM) that allows multiple operatingystems to run concurrently on a host computer.
2 These are molecules that, when joined together, make up the structural unitsf RNA (Ribonucleic acid) and DNA (Deoxyribonucleic acid). RNA is one of the threeajor macromolecules (along with DNA and proteins) that are essential for all
nown forms of life. DNA is a nucleic acid that contains the genetic instructionssed in the development and functioning of most living organisms.
ion Management 34 (2014) 221–225
and hidden costs. Fig. 1 provides an illustration of how life sci-ence researchers can interact with an IaaS cloud infrastructure forprocessing and storing the massive amounts of data generated bythis type of research.
There is no doubt that Amazon is capturing a great deal of theHPC market, thanks to its Amazon Web Services (AWS), its cloudservice infrastructure, which includes Elastic Compute (EC2), itsmain IaaS platform. Google (with its new Google Compute Engine)has recently entered the IaaS market and analysts are already seeingit as a potential and major threat to Amazon in this field (Darrow,2013). Pharmaceutical and medical research companies and insti-tutes, especially those engaged in genomics research, count amongAmazon’s cloud customers. For example, in the US, the Broad Insti-tute of MIT and Harvard along with Harvard Medical School areusers. Life Technologies, an instrument maker, and Seattle-basedGeospiza, a bioinformatics software company, have a partnershipto use Amazon’s servers to store genomic data. DNAnexus (a Cali-fornian bioinformatics startup), has built its business model aroundusing Amazon Web Services (Timmerman, 2010).
However, an increasing number of small ICT companies(especially those servicing the life sciences sector) are enteringthe cloud market. Among those are GenoLogics, GenomeQuestand Geospiza. For example, GenoLogics focuses its collaborativedata-management software platform on biomedical and drug dis-covery/development applications in the cloud, with emphasis ontranslational medicine and systems biology in pharmaceutical,biotechnology companies, and academic organizations. Genome-Quest provides a SaaS solution that allows researchers to performsequence alignment and data mining on next-generation sequenc-ing data. Sequence alignment is a method of comparing two or morenucleotide or protein sequences in order to determine the degree ofsimilarity between them that might be interpreted as the result offunctional or evolutionary relationships between them. The solu-tion provides centralized sequence data management resourcesand applications for biological research and other tasks such aspatent research. Geospiza offers a solution called “GeneSifter” fornext-generation DNA sequencing in the cloud through AmazonWeb Services.
Given the potential of cloud-based genomic research, acqui-sitions of small innovative companies in this area by largepharmaceutical, biotechnology or ICT companies is likely to accel-erate. Recently, Geospiza was bought by Perkin Elmer, the giantmaker of tools for life scientists, and CloudSwitch was bought byVerizon, a major global provider of communications services andan emerging cloud vendor.
The highly data-driven and integrative nature of research ingenomic medicine presents significant challenges in formulatingand testing important translational hypotheses. Advances in highthroughput experimental technologies continue to add to the expo-nential growth in publicly available genomic data. The integrationand interpretation of these immense volumes of data toward direct,measureable improvements in patient health and clinical outcomesis a great challenge in genomic medicine due to the scale of the com-putation and the amount of processing power required (Dudley,Pouliot, Chen, Morgan, & Butte, 2010). On that basis, according toDudley et al. (2010), cloud computing can potentially offer an effi-
cient and economic means to facilitate this task. In a case studydesigned to use statistical analysis to discover cancer-associatedeQTLs3 through integration of two high-dimensional genomic data
3 eQTLs stands for ‘expression quantitative trait loci’. They are genomic loci (i.e.,specific locations of a gene or DNA sequence on a chromosome) that regulate expres-sion levels of mRNAs or proteins. A gene expression is the most fundamental levelat which the genotype (the genetic makeup of a cell, an organism, or an individ-ual) gives rise to the phenotype (an organism’s observable characteristics or traits
N. Sultan / International Journal of Information Management 34 (2014) 221–225 223
sts, da
tEpbwc
ssChdiwtitirtuyondtmago
4
ufmsq(tr
sb
Fig. 1. Data flows between scienti
ypes (gene expression and genotype), the authors used Amazon’sC2 to provision one hundred virtual server instances in order torovide the computational power required for the experiment’s 13illion distinct statistical computations. By doing so, the authorsere able to demonstrate the viability and economy of using cloud
omputing for such HPC tasks.In June 2011 Syapse, a US biology-oriented research and
oftware developer, announced the adoption of its cloud-basedoftware, Biomedical Development Management (BDM), by theanary Center for Cancer Early Detection at Stanford in order toelp with accelerating translational research aimed at early canceretection. The company was spun out of Stanford University and
s currently led by a multidisciplinary team of entrepreneurs, soft-are developers, and scientists (HPC in the Cloud, 2011). One of
he most interesting Corporate Social Responsibility (CSR) projectsn the “novel” ICT-enabled fight against cancer is Dell’s “Poweringhe Possible” project which is aimed at donating cloud comput-ng infrastructure (as well as provide funding and fund-raisingesources, and IT staff support) to global research endeavors thatarget pediatric types of cancer such as neuroblastoma. This partic-lar pediatric cancer is rare and deadly and strikes one in 100,000oung children (usually before the age of 5) and is responsible forne in seven pediatric cancer deaths. It targets the sympatheticervous system, which controls heart rate, blood pressure andigestion, and gives rise to aggressive tumors that are often uniqueo each child. Dell’s cloud computing resources are expected to dra-
atically reduce the amount of time needed for genomic mappingnd analysis of patients’ tumors, enable early identification of tar-eted cancer treatments and create a repository of real-time datan pediatric cancer research by scientists (Dell, 2011).
. Coping with cloud concerns
The cloud IT service delivery model has the potential to benefitsers in many respects. One of its most significant is its ability toree users from the expense associated with buying, installing and
aintaining applications and hardware locally. The cloud modelhifts much of this expense to a pay-as-you-go model and conse-uently offers significant cost advantages according to one view
Lin et al., 2009). Moreover, the cloud reduces power consump-ion. A great proportion of the costs of running an IT infrastructureelates to electricity use which is essential for running hardware
uch as its morphology, development, biochemical or physiological properties, andehavior.
ta repository and a cloud system.
(e.g., PCs, servers, switches, backup drives) and cooling which isalso needed to reduce the heating generated by the hardware. Thisfactor could have huge implications for the environment. Users ofcloud computing will be able to significantly reduce their carbonfootprint at a time when government and regional regulations (e.g.,UK’s Carbon Reduction commitment and EU Energy Using Prod-ucts Directive) are putting pressure on organizations to reduce theircarbon footprints.
Notwithstanding these advantages, there are currently threemajor concerns that inhibit many organizations from using cloudcomputing; notably: security, interoperability and outages. Earlyand recent surveys indicate that security and availability (i.e.,non-interruption of service) are rated highly by organizations con-templating the adoption of cloud computing (see Cisco, 2009;Goldstein, 2009; North Bridge, 2013).
Vendor-lock is currently one of the main issues that cloudproviders try (but might find difficult) to resolve. Many cloudproviders offer their services through proprietary APIs (Applica-tion Programming Interfaces). This means that users of one cloudprovider may not be able to switch easily to another provider ifthey decided they wanted to due to poor service by the originalprovider. The cloud may have a long way to go before becoming autility such as that of water or electricity whose suppliers can bechanged without too much disruption to the users.
The third main concern is availability (or non-interruption ofservice). Outages are often the main causes of service interruption.Outages dating from 2008 to 2013 disrupted the services of manyof the big cloud providers such as Salesforce.com, Amazon, Googleand Microsoft (see McCarthy, 2012; Raphael, 2011, 2013).
Cloud computing may not be suitable for all organizations. Forexample, for large companies (especially those engaged in onlineretailing), the loss of service as a result of cloud glitches wouldbe a major concern, particularly if it impacts on their customersand results in substantial loss of sale opportunities and customerdissatisfaction. The issue of outages with relation to cloud ser-vices will continue to be a problem. However, for small companiesstruggling to survive the current global economic downturn andcash-strapped educational establishments, often used to similarglitches caused by their old in-house systems, cloud computing islikely to remain an attractive option due to its cost structure andflexibility.
For organizations involved in scientific and medical research,
most of the aforementioned concerns may not be as important tothem as they might be to those who provide products and ser-vices to consumers e.g., e-commerce companies. For example, theloss of a few hours of services may not be as dramatic for an
2 ormat
ooaefcctsci
oaiiatta
pnRubastpLc
S
ats(
5
ptiscvgaims
24 N. Sultan / International Journal of Inf
rganization conducting a research experiment as it would be fornline auctions or online retailers. Furthermore, issues of privacynd data protection are likely to be of less concern or indeed rel-vance, especially if research organizations only use an IaaS cloudor high speed compute operations. So, while the aforementionedloud drawbacks will remain of concern to many organizationsontemplating the use of cloud computing, their current advan-ages are likely to outweigh their potential disadvantages for thecientific community. On that basis one could deduce that cloudomputing’s market share of HPC is more likely to continue risingn the years to come.
One of the greatest challenges, as explained above, for many lab-ratories setting up life science research has been the need to obtainnd maintain a computational infrastructure required for analyz-ng a vast flow of proteomics data generated by mass spectrometrynstruments used in determining the elemental composition as wells chemical structure of a molecule. Cloud computing providedhat ability by removing the high cost barriers (e.g., infrastruc-ural investments) and the complexities (e.g., learning, training)ssociated with such undertaking.
The advantages of using cloud computing for other scientificurposes were illustrated by researchers working on the STAR4
uclear physics experiment at Brookhaven National Laboratory’selativistic Heavy-Ion Collider (RHIC). The researchers needed sim-lation results to present at the Quark Matter physics conferenceut found it difficult to do so as all the computational resourcest their disposal were either committed to other tasks or did notupport the environment needed for STAR computations. However,he STAR researchers were finally able to use an open source IaaSlatform (called Nimbus) developed by the US’s Argonne Nationalaboratory in order to dynamically provision virtual clusters onommercial cloud computers and run the additional computations.
Jerome Lauret, software and computing project leader for theTAR project, commented:
The benefits of virtualization were clear to us early on. . .We canconfigure the virtual machine image exactly to our needs andhave a fully validated experimental software stack ready for use.The image can then be overlaid on top of remote resources usinginfrastructure such as Nimbus (Argonne National Laboratory,2009).
Commenting further on the flexibility of cloud computing, hedded that, thanks to cloud computing, ‘a 100-node STAR clus-er can be online in minutes [whereas] Grid resources available atites not expressly dedicated to STAR can take months to configure’Argonne National Laboratory, 2009; see also Sultan, 2010)
. Conclusion
Using cloud computing for HPC purposes was not, until recently,erceived (by either users or providers) as a viable option due tohe performance overhead associated with the process of virtual-zation (one of the major technologies underpinning this new ITervice). However, such technology-related issues have been over-ome by developments in new hypervisors (the core technology ofirtualization) as was demonstrated in this article. To highlight therowing popularity of cloud computing for HPC (especially in therea of life science research) a number of examples were provided
n this article which demonstrate the potential of this ICT service
odel, not only in reducing costs but, most importantly, facilitatingcientific research.
4 STAR stands for Solenoidal Tracker at RHIC.
ion Management 34 (2014) 221–225
There are still major issues with the cloud model that relateto security, interoperability and availability which prevent manyorganizations from embracing this IT service model. For some orga-nizations (e.g., those engaged in online retailing) these issues areunderstandably important and they explain why surveys ratedthem among the main concerns that organizations have whendeciding to consider a cloud implementation. However, for someorganizations (e.g., those engaged in scientific research), as arguedin this article, such concerns are more likely to be of less impor-tance. These organizations might find that cloud computing’spotential disadvantages are more likely to be outweighed by theircurrent advantages. The potential of using cloud computing inadvancing development in life science research offers many excit-ing possibilities for professionals and organizations working in thisfield and could be key in accelerating the quest of finding cures tohumanity’s major illnesses.
References
Argonne National Laboratory. (2009). Nimbus and cloud computing meet STARproduction demands. http://www.hpcwire.com/offthewire/Nimbus-and-Cloud-Computing-Meet-STAR-Production-Demands-42354742.html?viewAll=y.Accessed 05.04.10
Baines, T. S., Lightfoot, H. W., Benedettini, O., & Kay, J. M. (2009). The servitizationof manufacturing: A review of literature and reflection on future challenges.Journal of Manufacturing Technology Management, 20(5), 547–567.
Cisco. (2009). The Cisco Powered Network Cloud: An Exciting Managed ServicesOpportunity. White Paper. http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/white paper c11-532553.html. Accessed 23.09.10.
Darrow, B. (2013). Amazon is the cloud to beat, but Google has the cloudto watch. Here’s why. http://gigaom.com/2013/04/02/amazon-is-the-cloud-to-beat-but-google-has-the-cloud-to-watch-heres-why/. Accessed 18.12.13
Dell. (2011). Dell Commits Funding, Cloud Technology and Employee Engagement toFight Neuroblastoma and Other Pediatric Cancers. http://content.dell.com/us/en/corp/d/campaigns/dell-healthcare-giving#campaignTab2. Accessed 04.12.11
Dudley, J. T., Pouliot, Y., Chen, R., Morgan, A. A., & Butte, A. J. (2010). Transla-tional bioinformatics in the cloud: An affordable alternative. Genome Medicine,.http://genomemedicine.com/content/2/8/51. Accessed 02.12.11
Goldstein, P. (2009). Alternative IT Sourcing Strategies: From the Campus to theCloud, EDUCAUSE Center for Applied Research. http://net.educause.edu/ir/library/pdf/EKF/EKF0905.pdf. Accessed 04.12.11
HPC in the Cloud. (2011). Canary Center at Stanford Adopts Cloud-BasedSyapse to Accelerate Cancer Research. http://www.hpcinthecloud.com/hpccloud/2011-07-19/canary center at stanford adopts cloud-based syapse toaccelerate cancer research.html. Accessed 02.12.11
Lin, G., Fu, D., Zhu, J., & Dasmalchi, G. (2009). Cloud Computing: IT as a Service. ITProfessional, 11(2), 10–13.
McCarthy, J. (2012). Down For The Count: 9 High-Profile Cloud Outages. CRN.http://www.crn.com/slide-shows/cloud/240012554/down-for-the-count-9-high-profile-cloud-outages.htm?pgno=4. Accessed 21.12.12
Niccolai, J. (2009). Penguin puts high-performance computing in the cloud. InfoWorld. http://www.infoworld.com/d/cloud-computing/penguin-puts-high-performance-computing-in-cloud-282. Accessed 05.12.11
North Bridge. (2013). 2013 Future of Cloud Computing Survey Reveals Busi-ness Driving Cloud Adoption in Everything as a Service Era; IT InvestingHeavily to Catch up and Support Consumers Graduating from BYOD to BYOC.http://www.northbridge.com/2013-future-cloud-computing-survey-reveals-business-driving-cloud-adoption-everything-service-era-it. Accessed 18.12.13
Raphael, J. R. (2011). The 10 worst cloud outages (and what we can learn from them).http://www.infoworld.com/d/cloud-computing/the-10-worst-cloud-outages-and-what-we-can-learn-them-902?page=0,2. Accessed 03.12.11
Raphael, J. R. (2013). The worst cloud outages of 2013 (so far). InfoWorld.http://www.infoworld.com/slideshow/107783/the-worst-cloud-outages-of-2013-so-far-221831#slide8. Accessed 18.12.13
Reid, C. (2008). The genomic frontier: Personalized medicine in action.http://venturebeat.com/2008/11/18/the-genomic-frontier-personalized-medicine-in-action/. Accessed 04.12.11
Rubenstein, K. (2010). Cloud Computing in Life Sciences R&D. http://www.insightpharmareports.com/uploadedFiles/Reports/Reports/Cloud Computing/SamplePages.pdf. Accessed 04.12.11
Shainer, G., Sparks, B., Schultz, S., Lantz, E., Liu, W., Liu, T., et al. (2010).Cloud Computing Will Usher in a New Era of Science Discovery. HPCWire. http://www.hpcwire.com/specialfeatures/cloud computing/features/Cloud-Computing-Will-Usher-in-a-New-Era-of-Science-Discovery-82710127.
html. Accessed 10.04.10
Sultan, N., & Salim, S. (2010). Managing IT resources the smart way: The case forvirtualization. International Journal of Computer Research, 18(1), 65–80.
Sultan, N. (2010). What can cloud computing offer to scientific research? Interna-tional Journal of Computer Research, 18(1), 49–63.
immerman, L. (2010). Amazon’s Cloud Computing Service Sees Opportunity inGenomic Data Overload. XCONOMY. http://www.xconomy.com/national/2010/07/06/amazon-with-rented-server-space-in-the-cloud-sees-opportunity-in-genomic-data-overload/. Accessed 01.12.11
andermerwe, S., & Rada, J. (1988). Servitization of business: Adding value by addingServices. European Management Journal, 6(4), 314–324.
argo, S. L., & Lusch, R. F. (2006). Service-dominant logic: What it is, What it is not,
What it Might be. In R. F. Lusch, & S. L. Vargo (Eds.), The service-dominant logic ofmarketing: Dialog, debate, and directions (pp. 43–56). Armonk, NY: ME Sharpe.
argo, S. L., & Lusch, R. F. (2008). From goods to service(s): Divergencesand convergences of logics. Industrial Marketing Management, 37(3),254–259.
ion Management 34 (2014) 221–225 225
Dr. Nabil Ahmed Sultan is Acting Head of the School of Business, Leadershipand Enterprise and Head of Division of Management, Business and Enterprise atUniversity Campus Suffolk (Ipswich, UK). Prior to that he was Award Director ofInternational MBA at Liverpool Hope University’s Business School. He is also Visit-ing Professor at the Servant Leadership Centre for Research and Education (SERVUS)at Vrije University, Amsterdam (The Netherlands). Dr. Sultan has a colourful profes-sional career and research background. He spent his early working years in the Arab
Gulf region and later headed a UK business. He also worked for the UNDP in Adenand New York before moving into academia in the late 1990s working initially atthe University of Liverpool. He has a research interest in information management,cloud computing, leadership, ethics and the socio-economic development in theArabian Peninsula.
