Services and Types of Cloud Computing - Case Study Example

Full Paper Introduction Cloud computing is all about making effective use of the computing assets i.e. hardware and software which are served jointly usually over (internet) network. We refer it as ‘Cloud Computing’ because it holds the symbol of Cloud which usually symbolizes huge networks particularly the ‘Internet’. The obstacles in communication associated with the system diagram are demonstrated by the ‘Cloud’ symbol. Cloud computing utilizes the inaccessible services to handle customer’s data, software and working out for the requirements and mitigates the user in managing these prerequisites (OBrien, J., Marakas, G., 2010). Monetarily, the core advantage of cloud computing is that the user goes for the services they really required and not get stressed about hit the highest point services, the user is actually assumed to pay for what he has been utilizing only, rather than long calculations and getting upset for prospects. The ease of access of the resources is offered every time through the internet from the Cloud. There is a stress free time for users as they really don’t need to deal with the fixations at the other end. The customers only acquire the particular IT services they actually need just the way they would go for any other service. Due to this additional advantage, Cloud computing has also been acknowledged as ‘Utility Computing’. In a world of computing, where considerable resources along with the huge expansion proposal is ever more becoming a massive venture for the IT industry, the perception of a highly scalable group of hardware and software resources has appeared as a sparkling star. An organization gets to spend much lesser and have less investments by opting for Cloud services as it gives an additional advantage to the cloud users by following the rule of ‘pay as you go ’which encourages the firms to cut off their expenses efficiently. The actuality now is that the organizations are no more liable to pay unnecessarily for the services they does not utilizes, in order to meet high practice and facilitates the user to pay the actual services acquired which enables them to save an enormous portion of money for them. Cloud computing is referred as a technology designed to smoothing the process of suitable network contact to a mutual compilation of constructive computing resources (i.e. networks, servers, physical and virtual storage devices, applications and services) which can be hastily presented and commenced with the slightest amount of management exertion or service source interface. Following are the obligatory provisions for Cloud Computing. These strictures should be accomplished so that it at accurately reckons any service as being Cloud based and lacking them being marked the true cloud computing technology would not be executed (Nowduri, 2011). Self -Servicing A customer who requires computing services immediately, at a definite time, can be grated with the needed resources ( like CPU-time, network storage, software, virtual storage, development platform etc.) in an involuntary (i.e. well-situated, self-conditioned and measurable) manner without the need of having the call for any human interaction with source providers. Network Access The essential computing resources are mostly offered over internet (network) and several customers utilize them (applications) which possess diversified proposal. Like cellular phones, microcomputers which include laptops and PDA’s situated at the user’s end .The excessive value is provided by the broad network access and it also boosts the capacity of the advantages provided by Cloud Computing. Resource Pooling Cloud service providers cater computing resources in a grouped form (i.e. collected together) so that various customers could be served. This could be attained by means of “multi-tenancy” focusing several customers of the similar resources or by virtualization focusing virtual devices indicating physical hardware. This is achievable by possessing various physical resources along with virtual resources with dynamism in order to allocate and reallocate according to the customer’s needs. Economies of scale and specialization of resources plays a role of the ‘Motivational Forces’ following the idea of a pool focused Computing theory. The outcome of such a shared model is the gloom of physical computing resources. This indicates that the resources are concealed from the users, the customer does not possess any know how of the location, structure and foundation of the resources which they have acquired. E.g. Users are unable to locate their data that where it is assumed to be stored in a definite Cloud. This facilitates a genuine user-friendly condition of resources to the customers and consequently allows resource grouping without sacrificing the secrecy of management structure of the resource provisioning methodology. Services and Types of Cloud Computing We can describe Cloud computing by classifying it into two categories. There are generally two ways of categorizing the clouds. First one produces from the several services provided by the Clouds while the other one produces from the types of the Clouds based upon the magnitude and user admission system demonstrated by the Clouds. Types of Cloud Clouds can be alienated into three types based basically on the network size, the customer’s number and the protection of admission. The types are (Bento, A. M., & Aggarwal, A. 2013): Public Clouds, Private Clouds, Hybrid Clouds Public Clouds Public Clouds illustrates the conventional meaning of Cloud computing technology. They symbolize measurable, animatedly available usually virtualized resources communal over the network (Internet). The offsite third party source measure resources and charges its users on the pay-as – you go basis. E.g. Think Grid, an organization caters a multitenant architecture for providing services which involves software –as-a service, platform-as-a-service, hosted storage etc. Other famous Cloud sources include Sales force and Amazon EC2. Private Cloud Private cloud is a word that describes a computing architecture providing hosted services on private networks. Outsized organizations opt for this type of Cloud computing to offer in-house services for consumers which are contained by the organization. The company network and its information center managers can create a local cloud for their own consumers. On the other hand, private clouds abolish several advantages associated with Cloud computing, primarily because firms are required to procure, constitute and control their personal clouds and therefore are unable to gain the proper advantages of Cloud computing as far as the commitment-less flexible accessibility of resources is concerned. Hybrid Cloud Hybrid Cloud as its name describes is a blend of resources from the internal and external resources, yet the most dependable and prominent option for business and firms to go for. E.g. an organization could have a choice in using a Public Cloud service for computing at daily basis but the business related information would be concealed in its own data center employed by the firms. Highly regarded and huge scale organizations have already invested a big capital in its communications needed to offer resources in-house; this could turn out to be one of the valid motives for the firms to acquire a Hybrid Cloud. The other reason could be the security associated with the public Clouds. In fact, the public cloud being deficient in providing the information secrecy has played as a key barrier in the use of Public Clouds (Ryan, 2011), the data security provided by the Public Clouds has forced the organizations to go for the Private Clouds as far as their data authenticity is concerned. Virtual Private Cloud A focused type of Cloud is the cryptographic Cloud storage launched by the mind bogglers at Microsoft. This specific storage solution facilitates the consumers with the security delivered by a Private Cloud while ensuring expenditure reserves intrinsic of a Public Cloud. Cloud Services The Cloud has been distinguished on the foundation of three unusual service models according to various individuals, according to our assumptions all the three basic services of the Clouds are Hierarchical in such a manner that one presents the base for the others. The service models are defined as: Software as Services (SaaS): In this service model, the Cloud users comprehends their relevant applications on a hosting environment which can be monitored through diversified network based consumers by different customers. The consumers may be the Web browsers, PDAs, desktops etc. This service cloud model has been used most extensively but on the other hand it is the most misinterpreted one as well, and consequently a lot of perplexity and argument has come forth in relation to this service model. Open Stack Open-Source Offerings: In a while, the discussion on the ‘Open Nebula and Open Stack’ will in depth listing of the corresponding functionality which is mentioned in the next section of this paper. Open Nebula is referred as an Open-source IaaS Solution which was primarily nurtured within the structure of an EU-funded project. Open Nebula organizes storage, network, virtualization, monitoring and security technologies to facilitate the forceful assignment of VM’s combining both information center resources and inaccessible cloud resources, as per provision strategies (Kerner, 2013). It possesses a huge customer base with telecommunication firms, system incorporators, and supercomputing centers. Open Stack is an open-source IaaS provider venture, which obtains sustainability by multiple cloud service providers. The venture was originated by NASA and Rackspace in 2010, whereas at this time various firms have become its component (e.g. HP, Cisco, Dell etc.) (DATA MONITOR: Dell inc. 2009). Two sub-projects are included along with the terms of the Apache License for its discharge. Nova for computing resources (virtual machines, network, block storage) (Kerner, 2013). A crucial goal of the Nova Project is to give backing to various IaaS APIs at the matching time. At present two APIs are being executed: Open Stack API 1.0 (corresponding the present Rackspace Cloud Servers API) and the Amazon EC2API. Swift for file- based storage preference. Earlier, the Swift codebase used to be an open-sourced edition of the engine behind Rackspace’s Cloud Files service. Nova began as an incorporation attempt between the NASA Nebula IaaS codebase and Rackspace’s Cloud servers API. Eucalyptus (Elastic Utility Computing Architecture for Linking Your Programs to Useful Systems) is considered as an open-source resolution (under the GNU General Public License) offering IaaS. It incorporates Amazon’s Public Cloud EC2 with other public clouds intended to pursue. The virtual machines that are controlled are XEN- OR KVM-based Linux System. Furthermore, there is compatibility for several VMware technologies in the (closed source) business version. Consumers can easily upload custom images as Eucalyptus does not contain public images. Considering storage, the Eucalyptus Block Storage offers a platform that is supported by the elastic Block Storage from Amazon. It facilitates the consumer to produce masses, include them in a virtual machine and to take prints of them, at the same time a storage manager known as Walrus is an execution of the Amazon S3 edge. As there is not any build-in controlling system, conversely auto scaling is not compatible, however the Eucalyptus works in align with Nagios and Ganglia. It should be kept in mind that as per , controlling and scheduling are intrinsic cloud distinctiveness therefore; any open source IaaS solution must involve these elements inside its foundation architecture (e.g. Open Stack provides Swift coverage to Statsd/ Graphite. Open Nebula and Open Stack In this part, we estimate two open-source IaaS solutions considering qualitative and a quantitative scenarios both3.1 Testbed Description assuming for our testing and evaluation, we have utilized two (2) physical hosts (PH) with the subsequent (similar) hardware attributes so that the evaluation of the IaaS approaches on peak of the same physical hosts could be performed. We have established OpenNebula v3.3.0 on the first physical host and OpenStack (Nova) v2012.1 on the other one, on these physical hosts. As far as the VMs configuration is concerned, it’s shown in the following table. Furthermore, the configuration of the VCPU is kept equal as a common decisive factor for our evaluation GRIFFIN, J., & DANSON, F. (2012). Qualitative Comparison: We discussed previously that the criteria assumed in our qualitative study that have been used to evaluate OpenNebula and OpenStack. Criteria differ from sustained functionality to security and are as pursue: Models Compatibility: It symbolizes the operational models (i.e. public, private, hybrid, community) which might show compatibility to OpenNebula and OpenStack. Virtualization compatibility: this entails to the compatible fundamental virtualization technology along with the other elements like whether reconfiguration that can be done in the runtime, with or without rebooting, allocation of the virtual resources, etc. Access: It demonstrates the APIs allowing administration of virtual machines with out-of-band way in as used in several cases throughout debugging. Image Support: This criterion will cover the compatible operating systems and whether a consumer can upload custom operating system images with already installed system, will be evaluated. Resource Selection: It covers the capability to choose the geographical location of the physical resources on which the virtual ones will be installed. Storage Support: It demonstrates the availability of the block storage or file storage and if a Content Distribution Network (CDN) is provided. Auto-scaling Support: It gives exposure to the ability of auto-scaling following accessing data (e.g. CPU or memory usage). High-availability Support: If the sources facilitate consumers to acquire high-availability involving the use of resiliency mechanisms (e.g. live migration, fault- tolerance, automatic backups, etc.). Monitoring Capability: Consumers may find information and notifications about performance deprivation, hardware failures, etc. API Support: Edges provided by the IaaS sources based on many protocols (e.g. SOAP, REST, etc.), API descriptors (e.g. WSDL), standards supported, etc. Security Features: In this criterion, right to use control management, certificates, password restrictions, source side firewall, etc. are included. Service Level Agreements (SLA) capability: facilitating consumers to negotiate authorizes and scrutinize SLAs with definite aspects. We additionally confer the qualitative results of the comparison between OpenNebula and OpenStack in Section 3.4 (“Results Evaluation”). 3.3 Quantitative Comparison: In this section we presented a quantitative comparison of the proposed open-source IaaS solutions. The aim of this comparative study is to evaluate the offered IaaS Solutions from the view of an End-user, i.e. without having any “internal” know how, the user tends to follow processes e.g. during the startup of the VMs. We observed through qualitative analysis and qualitative evaluation that there are some functional features which are not available in OpenStake but OpenNebula provide provides that features and make it more user friendly. For VMs deployment, OpenNebula has a web interface while OpenStake uses command line interface for this purpose. Along with this, web interface of OpenNebula provides the feature of resource selection for VMs deployment and also OpenStake does not support some features like auto-scaling and high availability support. One of the interesting points about both OpenStake and OpenNebula is that for SLA negation and management these two do not provide a native mechanism (which is available in ecosystem tools). Quantitative results shows that by allowing auto-scaling feature in OpenStake and purposeful storage support in OpenNebula would be beneficial for both respectively. From assessable point of view, OpenStake has a better performance than OpenNebula when scaling out (increment in number of VMs) during the deployment process as shown by the similar pending time. In spite of that, in massive at-once (not stepped) VMs deployment case, OpenStake has better performance than OpenNebula because OpenStake takes less time in getting the VMs running state from the acting state. Moreover, using quantitative results of at-once massive VMs deployment case, OpenNebula can get benefit is such algorithm is implement which decreases the under deployment VMs pending time. Quantitative and qualitative analysis evaluation results show that for the applications which require quick resource scaling (number of increment in VMs causes stepped deployment of new VMs 75% faster in OpenStake than OpenNebula ), OpenStake is the optimum solution . These applications may be called as soft real time applications (such as Virtual and Augmented Reality, Digital Film Postproduction, etc.) and as the applications of depicting spikes in requested resources-usually those applications which are open for a large group of people (e.g. Animoto application on Facebook) (Kerner, 2013). While OpenNebula would be more suitable for the applications having high elasticity requirements (this feature is not supported in OpenStake) such as smart city scenarios (e.g. by on-demand obtaining and analyzing information for enhancement in car cruise from city sensors smart phones and other cars, etc) or in scenarios of healthcare which uses OpenNebula features of security features. OpenStake Security In this section the OpenStake, its component and mechanism will be discussed. The publication of Open Stake the initial “austin” release in October, 2010 was consisted of two projects: Compute and Object storage. Object storage was ready for production while compute for testing. An updated version of OpenStake named as “bexar” was the release in February 2010. This updated version includes a new component “OpenStake Image service”. Moreover, enhancement was made in previous projects by development team. For Example, introduction of “swAuth” for authorization and authentication in object storage (swift) project. Then a third version was released with name “cactus” with two additional features to object storage project i.e., serve static content option and content checksum validation ability during get object actions. There was enhancement in performance of OpenStake in providing virtualization technology. The fourth version was released in September 2011 and named as “diablo”, at this point the community of OpenStake increased over 1500 people and 87 companies and also increases the product development. Nevertheless, improvement was made in scalability, availability and stability but only security concerns need to be considered now. OpenStack is open-source software for building private and public clouds. OpenStake includes three main projects. The key services were computing, storage, dashboard and networking, whereas supplementary services were identity and image. The OpenStake Compute (NOVA) developed by NASA manages and provides network of virtual machines. Infrastructure as a Service (IaaS) is offered by public cloud service while within Organization offered by private clouds. Hadoop and High-Performance Computing (HPC) are the tools used for computing OpenStake. OpenStake Compute consists of following: Management of Commodity Servers (including CPU, memory, disk and network interfaces) Organization of Local Area Network (LAN) (including flat, flat DHCP, VLAN DHCP, ipv4 and ipv6 networks works) Virtual machine image management tools (i.e., importing, sharing and querying tools) Assignment (and reassignment) of Floating IP address to VMs Enhancement of VMs RackSpace efficiency by VM image caching on compute nodes which is and contributed to OpenStack object storage (swift and cinder). Objects and blocks for servers and applications are saved in OpenStake storage. Static data like machine images, backups and archives are stored in Object storage, which is uses distributed storage system to store data. OpenStake cloud saves these objects and files in disk drives which results in scalability and repeatability. OpenStake also provide block-level storage devices for computing tasks that require high performance storage, which is often required by databases, expandable file systems, or servers that access raw block-level storage. Following are the features of OpenStake: Reduction in per byte cost of storage in commodity hard drives. Highly redundant and reliable storage system as data are copied different sectors of the cloud. Large scale data storage capacity allowing billions of individual objects to save. Supports Amazon s3 (elastic block storage) API. Utilities like the management of account, container and storage monitoring features. OpenStack image repository (glance) This feature enables discovery, registration and delivery for disk and server images. For new users base image template can be created and administrator can construct and store images and can be saved in raw, VHD (Hyper-v) etc.. OpenStack networking (quantum): For cloud networks and IP addresses, OpenStack networking is an API-driven system. This feature includes: Management of Static, DHCP and floating IP addresses. Support of several networking models like Flat networks and VLANs. Creates and manages users. Support SDN technology (i.e., OpenFlow) OpenFlow (SDN) In system architecture, SDN stands for “Software-Defined Networking”. This is widely recognized but only its architecture is used more. Hardware and software both are components of network device. However, a network cannot be defined independent because of its API (Application Programming Interface). Hence OpenFlow is introduced as this can enable SDN. OpenFlow is not a networking method of providing Layer 2 or IP routing function but it only provides an interface for network device. The Open Network Foundation (ONF) has defined its specifications for promoting and standardization, allowing components and functionalities of switching and the OpenFlow protocol for managing OpenFlow switches from remote controllers. Accessibility and management of API hardware controlling is done by OpenFlow. However, device manufacturer keeps some information and it enables the users to manage the network independently. Many devices could be included in network frame of cloud like intrusion detection systems, load balancers and firewalls. OpenStack dashboard (Horizon) This feature enables the users and administrator to manage and control the computing, storage and networking resources. This also reduces the number of resources as it creates users and projects and assigns projects to users. Resource controlling and allocation is also done by OpenStake Dashboard. It is an extensible web-based application. OpenStack identity (keystone) (Kedar, S. 2009) It provides authentication and manages user database. A third party, back-end directory services, can be used for common authentication (i.e., lightweight directory access protocol or ldap).It can use various verification systems, such as the standard username and password, token-based systems and web services such as Amazon. It also allows administrator to develop policies among users and systems, create users and tenants and grant permission to compute, store and network resources. Security in OpenStack There are many flaws in OpenStake and thes can be address to the current released version of OpenStake. These are (Wilshusen, G. C. 2011): No password reset option to users on horizon; it can be reset only within administrator interface. The impact of this flaw in unknown. The project administrator is automatically made the administrator of whole system in horizon. For cloud computing, OpenStake uses the concept of projects and tenants to group people into logical units. However, administrator of a single project is assigned the managerial role of all the projects by the interface only. He gets the right of user and project creation, can even change projects and remove items. Network API uses Cleartext and OpenStack API endpoints encourage the use of cleartext but currently no SSL/TLS support is available. This allows middle attacker and even “sniffing” passwords over the wire can be trivial. No server-client system authentication allows any host having access to the db and to the AMQP system can act as a compute node and can launch VMs. Usernames and passwords Clear text stores the passwords and usernames for accessing images in db and external devices. While storing images on swift using glace, Clear text stores username and password of swift account along with URL of swift object in the bd. This may allow any cloud swift user to access information in db. This is an unnecessary storage of information as username and password is already stored in configuration file of glance. On basis of these problems, cloud security solution has studied which is discussed in later section. In prevision section of security in cloud computing, we discovered which issues are often discussed in relation to identity and access management. So this section discusses identity and access management. Risk Management Strategy Risk Management Practices of NASA’s were ineffective for Acquiring and Securing Public Cloud-Computing Services. According to NIST, it is a challenge to assess and manage the risk in transferring data to a public cloud as it is totally under the hand of the cloud provider. To overcome this, risk mitigation contracts should be develop, addressing the business and security risks unique to cloud-computing environments. Some clauses should be added in contracts with cloud providers explaining the measurement of performance, how it will be reported, and enforced and specify how Federal privacy, litigation discovery, and data retention and destruction requirements will be met. Moreover, it should be included in contract that how the cloud service provider will perform security tasks and it should be inspected by a third party. Finally, roles and responsibilities of the Agency should be clearly mentioned, the cloud provider, and the cloud broker must drive contractor performance and ensure Agency systems and data are adequately secured. For public cloud services specifications are generally known as service agreements or service contracts. This service contract includes all the terms and conditions for access and use of the services which the cloud service provider will offer, duration of service, conditions for termination, and disposition of data (e.g., preservation period) when the contract will terminate. Usually all these are contained in multiple documents, including a service level agreement and privacy and acceptable use of policies. NIST has defined two types of cloud computing service contracts: predefined, non-negotiable contracts and negotiated contracts. In first type of contract, the contract terms are prescribed by the cloud provider and these contracts do not impose requirements except the scope of service and availability nor do they address Federal IT security, privacy, data production, or retention and destruction requirements. Moreover, the cloud service provider keeps power to modify the contract without notifying the customer. Negotiated service contracts are like outsourcing contracts for IT services. Terms and condition can be tailored to address company’s requirements for tracking and reporting service( to check effectiveness), contains technical controls such as incident detection and handling, require compliance with laws and regulations, data ownership rights and the use of validated products having national or international standards. To adopt the best practice, the recommendation from the Federal Chief Information Officer (CIO) and Chief Acquisition Officer Councils is that contracts for cloud services clearly define how performance is guaranteed (such as response time, resolution or mitigation time, and availability), providers need to monitor their service levels and have to provide timely reports in case of any failure and if service levels are not met, there should be patently on service provider. On reviewing NASA’s contracts with their public cloud-computing services provider to check whether the Agency had implemented effective risk mitigation measures we find that contracts met best practices for attaining cloud services as recommended by the Federal CIO and Chief Acquisition Officer Councils, as well as practices identified in FedRAMP.12. For instance, we analyzed that contracts identifies the roles and responsibilities of the company and way to measure contractor performance, reporting procedure, and enforcement. We also evaluated that the contract addressed Federal privacy, discovery, and data retention and destruction requirements. Finally, it was stated that service providers or brokers appropriately addressed key IT security measures like incident detection and handling practices and service provider’s evaluating procedure. The summarization of results is given in Table. The five NASA cloud-computing contracts which were reviewed by us, three of them were for cloud services categorized as low-impact services and remaining two were for moderate-impact services. The three low-impacts and one of the moderate-impact service contracts were accepted by NASA. For the other moderate-impact cloud service, hosted by more than 100 internal and external Agency websites, NASA negotiated a contract with eTouch (a cloud-computing broker), to manage service delivery and helps ensure IT security and privacy requirements were met. Conclusion Open Stack is beneficial for companies looking for an edge in todays economy. Cloud computing services are offered by many vendors which is now getting affordable over time. Lower prices allow companies to use the staff for other available tasks and they can use their resources more efficiently in different projects and by paying for services only when needed. These features allow the companies to have a growing support for this model as it is attractive and uses economical pay-as-you-go approach. A major threat which is posed by Open Stack is that it obliterates internal and external boundaries in security concerns. There is a need to study cloud security in order to understand security of companies’ data. Second level is the availability, and suppliers may be victims of attacks that impede the functioning of its operations. For IaaS cloud environment, this work gives the comparison of two open-source clouding solution; OpenStake and OpenNebula. Both are Apache licensed released open-source software and developed over the internet using open and transparent process. OpenNebula provides private clouding for users while OpenStake provides AWS-like public clouding. Moreover, in OpenNebula functionalities can be offered using a single integrated bundle while in OpenStake a set of feature is compliment for individual functionalities and capabilities. The implementation of OpenNebula follows some standards like Amazon APIs, OGF, DMTF and SNIA specifications while the implementation of OpenStake follows the requirements of companies which are contributing in this and then forms an AWS. Qualitative and quantitative analysis can be summarized as; the OpenStack version has not much functionality (e.g. auto-scaling) but performs better than OpenNebula while implementing new VMs. In addition to the enhancements purposed in our paper, OpenNebula and OpenStake are optimum solution for different applications (e.g. in security-sensitive and city scenarios OpenNebula is preferable while in time-constrained scenarios OpenStake is preferable). In spite of this, the future evaluation of these open sources clouding solution will be dependent on their updated versions available to the community. These evaluations will also refer the hardware characteristics to identify that at which extent it will affect the provision of services. Moreover, we will perform more quantitative analyses of large infrastructure to collect additional information and will compare the IaaS open solution with Eucalytpus. References Bento, A. M., & Aggarwal, A. (2013). Cloud computing service and deployment models: Layers and management Business Science Reference. DATA MONITOR: Dell inc. (2009). ().Datamonitor Plc. GRIFFIN, J., & DANSON, F. (2012). Analytics and the cloud - the future is here. Financial Executive, 28(9), 97-98. Kerner, S. M. (2013). Open stack cloud goes global to draw in developers, technologies. EWeek, , 13-13. Kedar, S. (2009). Introduction to data processing In Database management systems (First ed., pp. 1). India: Technical Publications Pune.. Nowduri, S. (2011). Management information systems and business decision making: Review, analysis, and recommendations. Journal of Management & Marketing Research, 7, 1-8. OBrien, J., Marakas, G., (2010). Management Information Systems. 9th ed. New York McGraw Hill STAIR, R.M. and REYNOLDS, G.F., Principles of Information Systems (with Printed Access Card) Course Technology. Wilshusen, G. C. (2011). INFORMATION SECURITY: Additional guidance needed to address cloud computing concerns. GAO Reports, , 1. ZALESKI, J. n.d, Integrating Device Data into the Electronic Medical Record: A Developers Guide to Design and a Practitioners Guide to Application Wiley-VCH. Read More