European Journal of Scientific Research ISSN 1450-216X Vol.55 No.4 (2011), pp.569-579 EuroJournals Publishing, Inc. 2011 http://www.eurojournals.com/ejsr.htm A Cost-Effective Cloud Service for E-Learning Video on Demand Lavanya Rajendran Department of Media Sciences, Anna University, Chennai, India E-mail: lavanya@annauniv.edu Tel: +91-98400 09744 Ramachandran Veilumuthu Department of Information Science, and Technology Anna University, Chennai, India Abstract The main aim of this research paper is to develop a cost-effective model for storing and fetching e-learning Videos and to deploy the same in cloud environment, thus making e-learning Video on Demand as a Service more secure and scalable. The present study aims to develop an e-learning model in the Cloud Environment, which can be easily accessed by everyone without any constraints on geographical location. The video materials are stored in the cloud and through a progressive download approach and the REST API, the videos in the form of sequence of bytes have been streamed to the learner in real time. The e-learning VoD is developed similar to the real classroom environment, wherein a learner has to be present on time to get the complete video of the relevant course he /she has registered for. By providing a smooth and lively streaming of multimedia based e-learning materials, the learners are benefited with a cost-effective and Virtual Classroom environment. Comparison between the traditional VoD service and the proposed model shows that, the net savings for a particular case is approximately 30% of hardware and maintenance cost. The application is scalable and cost effective and therefore provides benefit to modern Educational Institutions in terms of investment and helps in smooth functioning of e-universities. Keywords: Cloud Computing, elearning, Video on Demand, Distributed Environment, REST API. 1. Introduction Education is defined as the conscious attempt to promote learning in others. The purpose of education is to develop knowledge, skill, and character of students. The education system which focused on direct teaching, now in the current scenario, has transformed into an education that focus on creating a viable and productive learning environment, regardless of however the teacher-centric environment might be. In the Traditional Auditory learning style, the learner s knowledge mostly depends on the oratorical skills of the teacher and in the Visual learning like powerpoint presentation, even though the learners were able to visualise the system, they were ignorant about how the system actually worked.
A Cost-Effective Cloud Service for E-Learning Video on Demand 570 To overcome these problems in traditional learning system, Mashup [1] and new innovations should be brought in educational systems like interactive country wide class rooms and simulations [2]. The elearning Video-on-Demand (VoD) learning system can offer learners additional approach to absorb new knowledge without the restriction of time and place. Many researches have proved that bandwidth is a key factor to commercialize VoD system. Ping Zhang et al proposed both centralized and distributed architectures of VoD learning system to reduce the network bandwidth. The challenging task in VoD applications is to satisfy diverse client requests for discrete videos with restrained resources [3]. Seethalakshmi et al has proposed a single-server/multi-client architecture that enables the clients to access the remote server at anytime for the required media. A Remote Method Invocation (RMI) based distributed model has been developed in such a way that for every request from the client for a media clip, the server marshalls the data to the client. During the last five years, e-learning courses were based on the Learning Management Systems in the browser. With new trends of Web 2.0 and e-learning 2.0, the e-learning developers have moved to Rich Internet Applications. The multimedia based e-learning materials stay as a backbone for several e-universities like IBA e-university, Asia e-university, UK e-university and several other e-universities. The UKEU (UK e-university) project was started with the idea of bringing the best of British Higher Education to students around the world. In 2003, only 900 out of 5000 students were expected to join this e-university, but unfortunately, the e-university was shut down due to lack of sufficient funds [4]. Similar to this project, there were several e-learning programs not picked up due to huge cost involved in delivering the e-learning contents to the learners. Even though, it is proved that simulations are increasingly being used as powerful and flexible educational devices [5], it is not yet fully successful because of the huge cost. It has been estimated that YouTube pays over 1 million dollars a month in bandwidth costs. These costs are expected to go up, as demand increases and higher-quality videos are made available [6]. The paper aims to develop a cost-effective e-learning Video on Demand as a Service model. It elaborates the workflow of the developed service and focuses on e-learning Video on Demand as a Service in cloud environment, thereby moving on to the new concept of Cloud Learning which will benefit the learners worldwide. Thus, this concept would lead to a fundamental shift in the education paradigm by providing a new way for storing and hosting the e-learning materials which gives a better opportunity for the learners to enhance their skills and to attain hands on experience in various fields. 2. The Cloud Architecture The term cloud computing implies access to remote computing services offered by third parties via a TCP/IP connection to the public Internet [7]. Cloud computing is a model for enabling convenient, ondemand network access to a shared pool of configurable computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction. This offers reliable services delivered through data centers that are built on computer and storage virtualization technologies. Therefore, it is a technology aiming to deliver on demand IT resources on a pay per use basis [8] and cloud uses the stateless protocol HTTP, to communicate with your computers [9]. Clients are the devices (mobiles, thick and thin clients) that the end users interact with, to manage their information on the cloud. The data center is the collection of servers where the subscribed application is housed. Distributed Servers are placed at geographically disparate locations. But to the cloud subscriber, these servers act as if they were buzzing away right next to each other in the cloud environment. The Cloud Computing Architectural model shown in Figure 1 offers different layered services, including Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS). Each segment serves a different purpose and offers different solutions to businesses and individuals around the world. Each of these services is self-describing and open components that support rapid, low-cost composition of distributed applications.
571 Lavanya Rajendran and Ramachandran Veilumuthu Figure 1: Cloud Computing Architectural Model. From the statistics taken among the users of cloud services, fifty one percent feel cloud is a convenient environment, forty one percent appreciate the ability to access their data from any computer, and thirty nine percent appreciate the ease with which information can be shared. (2008). 3. Proposed e-learning Video on Demand Model The work attempts to define a Cloud based model that makes e-learning Video on Demand as a Service in a cost-effective manner for both learner and the providers. In addition to cost effectiveness, it also enhances the security and scalability of the application. This model simulates the real time education system, where the lectures were given to the learners based on the time-table prepared. If a learner comes late to the class or if he fails to attend the period, then he would miss the class notes for that session. In the similar way, if the learner attends the cloud learning after the session starts, then he will be streamed with the same subject, same scene what the other learners of the same course are streamed with. This has been achieved through the concept of multicasting. Thus, unlike the traditional video on demand architecture which uses a dedicated channel to serve each and every request, thereby increasing the bandwidth requirement, the multicasting approach shares the single video among many users thereby reducing the cost per client and increasing the system scalability. Figure 2 shows the block diagram representation of the proposed Video on Demand model in the Cloud Environment. Figure 2: Proposed Cloud Based VOD model CLOUD SERVICE NETWORK STORAGE e-learning VoDaaS Video Blob Tables Client 1 Client 2 In the proposed model, the videos are stored in the centralized location in the cloud using the Storage as a Service layer shown in Figure 3. The server in the cloud environment has the control and can manage the videos stored in the cloud. It supports the streaming of all video file formats. The advantage with this model is that, even if thousands of requests are placed to the server at a time, the server will be able to stream those videos simultaneously to all the clients.
A Cost-Effective Cloud Service for E-Learning Video on Demand 572 Figure 3: Cloud Based Architectural Model for Video on Demand SERVER IaaS VIDEO BLOBS E- Learning VoD as a Service TABLES STORAGE AS A SERVICE WEBROLE CLIENT 1 CLIENT 2 CLIENT 3 The e-learning Video on Demand as a Service model is provided on top of an effective IaaS layer that manages virtualization of resources and multi-tenancy. The layer WebRole shown in Figure 3, is a single HTTP endpoint for the external clients. This role will access the table storage services via Storage Library and Representational State Transfer APIs. The service definition of this WebRole consists of a service name, the storage resources and the configuration settings and the values of them are configured when the WebRole instance is running. The service configuration for this WebRole includes number of instances to be deployed for each role and the values for the storage resources like account name and account shared key. When the number of clients increases, the multi-tenancy attribute of the server ensures for a scalable and secure environment. 4. Workflow For E-Learning Video On Demand From End User s Perspective The Workflow of the Cloud Service model for VODaaS from end user perspective is illustrated in Figure 4. When the learner enters the e-learning VoD, a new instance is initiated and the learner can basically do three different tasks, namely Register himself with the e-learning Courses, marked as 1, 1A or if he is a registered user, then, he can directly login to the e-learning VoD either as Teacher or Learner marked as events 2, 2A and 2B or directly watch the list of courses available marked as 3 in the Figure 4. The teachers are those who will upload or submit the e-learning Videos to the cloud storage. Learners are those who will access the videos for a stipulated time. Once the client logs in to the e-learning VoD as teacher, it is possible for him to upload or see the Videos marked as events 4, 5 and 6 in Figure 4. If the client is a learner, then he is given permission only to see the videos and he cannot upload e-learning videos. Before watching the e-learning video, the learner has to login and to confirm his registration for the course marked as 7 and 7A. Once he is logged in and confirms the registration, the requested course videos that are stored in the cloud storage are fetched to the learner as marked as events 8A and 8B. Depending upon the time at which the learner enters, three different tasks are carried out, that is marked as 9. The video is not streamed to those who arrive earlier than the Learning Session and those who arrive after the Learning Session marked as 10A. Those who arrive on time can watch the video from the beginning as marked as 10B. Those who come late, will be screened only from the portion that other learners are screened, which is marked as 10C and those who come early, have to wait till the session time to watch the e-learning Video.
573 Lavanya Rajendran and Ramachandran Veilumuthu Figure 4: Work Flow for E-Learning Video on Demand The progressive download approach is used to screen the e-learning video. It merges download and true streaming approaches. In other words, it simulates the true streaming and it is better than the download approach as it has many advantages of Live streaming. This approach is not expensive as it is streamed from the Web server. The Representational State Transfer (REST) API is used to upload the e-learning contents as blocks of 1MB size. The REST protocol is selected as it emphasizes on simple point to point communication over HTTP using plain old XML. The stateless characteristics of REST increases the security further more by accepting all the information required to understand the request submitted by the client each and every time. By uploading files through REST API, the application guarantees the successful uploading of files from any platform.
A Cost-Effective Cloud Service for E-Learning Video on Demand 574 It also ensures quicker and faster action. Adding REST components indicates the possibility of scalability. The following is the sample format of the REST API for PUT BLOCK REQUEST message (A), when uploading a 64MB size video to the blob storage service. The following Put Block request illustrates the uploading of a video file to the storage service in cloud by dividing it into small chunks of 1MB each. Those small chunks of video get committed only after receiving the PUT BLOCKLIST request (B). Sample Put Block Request PUT http://cloudeducation.blob.core.windows.net/computers/basicnetworking.mp4?comp=block&blockid= AAAAAA== HTTP/1.1 x-ms-date: Sat, 08 May 2010 02:44:03 GMT Authorization: SharedKey cloudeducation:a3zgr2hgibsidhi6vxaab6cgtuog0hbujgb8cbmka38= Host: cloudeducation.blob.core.windows.net Content-Length: 1048576 Expect: 100-continue Connection: Keep-Alive Put Blocklist Request PUT http://cloudeducation.blob.core.windows.net/computers/basicnetworking.mp4?comp=blocklist HTTP/1.1 x-ms-date: Sat, 08 May 2010 03:05:46 GMT Content-Type: text/xml; charset=utf-8 Authorization: SharedKey cloudeducation: EBGzSPgdcSHFBZakKk48o7gcYFZHrfRvMjvHqdWIDmI= Host: cloudeducation.blob.core.windows.net Content-Length: 1559 Expect: 100-continue <BlockList> <Block>AAAAAA==</Block><Block>AQAAAA==</Block><Block>AgAAAA==</Block> <Block>AwAAAA==</Block><Block>BAAAAA==</Block> <Block>BQAAAA==</Block>...<Block>QAAAAA==</Block> </BlockList> The above REST commands show the PUT BLOCK request and the PUT BLOCKLIST request for uploading 64MB file in the cloud storage and the following PUT BLOCKLIST response shows the response obtained after uploading. Put Blocklist Response HTTP/1.1 201 Created Transfer-Encoding: chunked Content-MD5: 5iuBQN7doG4fWc5vXAWLng== Last-Modified: Sat, 08 May 2010 03:06:09 GMT ETag: 0x8CCBC728AFEEE05 Server: Blob Service Version 1.0 Microsoft-HTTPAPI/2.0 x-ms-request-id: b69c08b0-a39c-4fd1-b330-6bdda8e1a8e6 Date: Sat, 08 May 2010 03:06:08 GMT From the above output, the HTTP status code 201 indicates that the PUT BLOCKLIST request was successful. Hence all 64 blocks of blobs each of 1MB size were committed, thereby uploading the complete video file in the Storage as a Service layer. The response contains the list of resource
575 Lavanya Rajendran and Ramachandran Veilumuthu characteristics like Server details, Content-MD5 and request id. The response also contains the ETag header field which indicates the current value of the entity tag that is now created for the given request. The ETag is used to provide cache validation. 5. Video Security Usually, in the client side, when the video is cached, there is a very high possibility that the video will be pirated. When e-learning Video on Demand as a Service is offered to larger audience, the caching is a problem and it is not feasible to support real time encryption with unicast connections. Hence to avoid piracy and to increase the security, the private key encryption is adopted. Depending on the level of authorization, the user is granted with one or more permissions to perform specific operations or actions. These actions typically map directly to important business functions, or to the management of the application itself. For example, a consumer can only read a video blob data to specified time duration that the provider permits. A provider can upload, read and delete a video data depending on his permission scope. Managing the access control is given based on the scope of the roles. Each scope inherits roles, permissions, and business rules from their parent. The following Figure 5 illustrates a sample access control provided to the clients. Figure 5: Authorization Access Control Users Roles Permission Business Rules Action Can Register for a video Client Request to watch a video If Registered for that video? Provider Can Appro ve for a video Can Reject for a video If within Time Period & has permission The signed access signature method is used to provide the read permission to the Video files for those who have registered for the course in a stipulated time. In this method, only after specifying the Private key of the service provider, the public key will be created, which is used by the learner to watch the e-learning contents for the stipulated time. The following URL depicts clearly that, only read permission is given for the basicnetworking.mp4 blob to the learner, which is specified as sr =b and si = readaccess. http://cloudeducation.blob.core.windows.net/computers/basicnetworking.mp4?sr=b&si=readac cess&sig=f3m37mro%2buws4d2%2flacdglfpyp%2baxjsfeugsm2ihamq%3d The unique string-to-sign denoted as sig is constructed from the fields like starting and ending time of the online classes, signed identifier, account name and Private key that are verified in order to authenticate the request. The signature obtained is an HMAC (Hash-based Message Authentication Code) computed over the string-to-sign and a public key is obtained by using the SHA256 algorithm (Secure Hash Algorithm), which is encoded using Base64 encoding scheme. 6. Video Fetching The Silverlight application, which was initially released as a video streaming plugins, now comes as a Web application framework that provides the functionality to integrate the video files into a single runtime environment. This player is used to play bytes of stream arriving at the clients end. The
A Cost-Effective Cloud Service for E-Learning Video on Demand 576 declarative programming paradigm XAML is used to execute the above task. XAML is an XML dialect which provides a way to bind presentational data and the declarative list of UI elements with some or all of the code used within them. This is a way to increase the processing power of the client as well as the server. When the student requests for an e-learning course at the correct stipulated time, then the whole video will be streamed to the system, else if the student arrives earlier, then he has to wait till the stipulated time. If the student arrives late, then he cannot view the whole video. If he comes half an hour late, then that half an hour video is not fetched for him. The users cannot pause, forward or rewind the video as the control is not given to them. Thus, all the students requested for an e-learning course will see the same material at a given time. This is executed with the following formula, where the Media Player Position or the video position is represented as MPpos, video Start time and End time are represented as MST, MET respectively. The Tc is the time when the client logs in. The TT represents the total length of the movie. MP pos = (M ET M ST ) * ((T c M ST ) (in Sec)) / (T T (in Sec) ) The following segment shows the sample HTTP GET request submitted to the server to access the video file. Get Request GET http://cloudeducation.blob.core.windows.net/computers/basicnetworking.mp4?sr=b&si=readaccess&si g=f3m37mro%2buws4d2%2flacdglfpyp%2baxjsfeugsm2ihamq%3d HTTP/1.1 Accept: */* UA-CPU: x86 Accept-Encoding: gzip, deflate User-Agent: Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.0; GTB6.4; SLCC1;.NET CLR 2.0.50727; Media Center PC 5.0;.NET CLR 3.5.21022; InfoPath.2;.NET CLR 3.5.30729;.NET CLR 3.0.30618) Host: cloudeducation.blob.core.windows.net Connection: Keep-Alive The HTTP response from the server for the above request is given below. The response 200 OK indicates that the e-learning video file which was requested can be viewed successfully. Only when the learner accesses the e-learning course at the given time, the specified video file will be unlocked. Response HTTP/1.1 200 OK Content-Length: 67582743 Content-Type: text/xml; charset=utf-8 Last-Modified: Sat, 08 May 2010 03:06:09 GMT ETag: 0x8CCBC728AFEEE05 Server: Windows-Azure-Blob/1.0 Microsoft-HTTPAPI/2.0 x-ms-request-id: 6b2ab78a-f777-4235-b632-af6b0d44230f x-ms-version: 2009-09-19 x-ms-lease-status: unlocked x-ms-blob-type: BlockBlob Date: Sat, 08 May 2010 03:12:09 GMT The developed e-learning Video on Demand as a Service in Cloud Environment offers scale up and scale down environment which is best suited for a scenario that does not know the number of clients accessing the video. This proposed model uses the multicasting approach, which does not use the dedicated channel for each and every request thereby reducing the cost and increasing the system scalability. In addition to it, this model is hosted on top of the IaaS, which also provides additional scalability and secure environment. To fetch the videos faster they are stored as BLOCK BLOBS that
577 Lavanya Rajendran and Ramachandran Veilumuthu are divided into 1MB block each. To avoid piracy, the videos are encrypted during runtime using the Private key. This model adapts to different video formats and can be delivered to different gadgets. 7. Cost Analysis The literature review shows that those organizations, which minimize the total operational cost and maximize the innovative investments have higher revenue and profitability performance. Through the proposed e-learning VODaaS model, the Universities can reduce the cost associated with hardware and maintenance required and thereby increasing their profit. The other advantage is that students from worldwide can study in any university, thereby gaining the advantages of International Education Standards. This will in turn increase the knowledge level of the learner. The comparison of the computing capacity and cost vs time between the traditional hosted video on demand model and the proposed e-learning VODaaS model is carried out. In the traditional hosted e-learning VoD model, the required hardware is installed in advance by forecasting the number of requests, the workload and the growth of the e-university. These service providers ensure that they invest on the hardware 40% to 60% more than the forecasted need. As the load increases, the capacity is also added ensuring over provisioning. The virtualization helps to decrease the over provisioning, but it has to be done manually by adding the required virtual machines or the physical server. Even by doing so, 20% of the capacity is over-provisioned, which not only increases the initial cost but also increases the maintenance cost. Inspite of these high investments, there are chances of unpredictable high growth in the load, which at times is very difficult to handle immediately, thereby causing poor performance in the server. There are also chances that the server might go down. This will surely reduce the client s trust on the service provider. The other disadvantage with the traditional model is that most of the time, the computing and the hardware capacities are not fully utilized, leading to excessive and needless investments. The computing capacity and cost incurred over time in the traditional video on demand model is shown in Figure 6. Here, the dotted line in blue color refers to the capacity forecasted, the dashed line in green color denotes the established, over provisioned capacity and the line in the red color denotes the actual computing capacity that has occurred. Figure 6: Computing Capacity and Cost analysis vs Time Traditional Video on Demand With the proposed e-learning VODaaS model, the concept of over provisioning can be drastically reduced by providing the capacity immediately when required and also by monitoring the loads dynamically. This model can reduce the capacity when the load reduces and dynamically increases to the required capacity when the load increases drastically. As the hardware utilization is fully automated, the idle resources can be used efficiently. The computing capacity and costs incurred
A Cost-Effective Cloud Service for E-Learning Video on Demand 578 over time for the proposed e-learning Video on Demand as a Service is shown in Figure 7. Here, the blue dotted line refers to the capacity forecasted. The dashed line in green color denotes the computing capacity provided by the proposed e-learning VODaaS, which is always excess to the actual computing capacity used, which is denoted by the line in red color. Figure 7: Computing Capacity and Cost analysis vs Time Proposed Video on Demand as a Service Comparing the computing capacity and cost vs time between the traditional video on demand service and the proposed model, the net savings with regard to the cloud environment for one particular case is approximately 20% to 40% of hardware and maintenance cost during the period of 30 months. The other advantages of providing e-learning Video on Demand as a Service model in Cloud Environment is that, the provider need not worry about the Storage requirement, Capacity Planning and Security Management. 8. Conclusion An enhanced, secured and cost-effective cloud computing architectural model for e-learning video on demand is developed and tested in the cloud platform. This proposed model ensures the costeffectiveness and scalability of the application. This new concept of Cloud Learning provides a smooth and lively streaming of e-learning materials to the students in a cost-effective manner. To provide high bandwidth possible to the clients, the Content Delivery Network(CDN) is enabled. When a client requests for a particular video, the video blob data is read directly from the Azure Storage. Instead, when a CDN is configured and a request is made using the CDN, the request is redirected to the CDN endpoint closest to the location from which the request was made to provide access to the video blob. The developed e-learning Video on Demand as a Service in Cloud Environment offers scale up and scale down environment which is best suited for a scenario that does not know the number of online learner who will be accessing the video. This proposed model uses the multicasting approach, which does not use the dedicated channel for each and every request thereby reducing the cost and increasing the system scalability. In addition to it, this model is hosted on top of the IaaS, which also provides additional scalability and secure environment. To fetch the videos faster they are stored as blobs that are divided into 1MB of blocks each. To avoid piracy, the videos are encrypted during runtime using the private key. This model adapts to different video formats and can be delivered to different gadgets. As the application is scalable, it provides benefit to the provider in terms of faster money back. Further, this model can also be utilized not only for educational purpose but also in the field of entertainment for providing movies online.
579 Lavanya Rajendran and Ramachandran Veilumuthu References [1] Benslimane D jamal; Schahram Dustdar, and Amit Sheth, Services Mashups: The New Generation of Web Applications (HTML). IEEE Internet Computing, 2008, 12, 5. [2] Chandola R.P, The real problems of Indian Education - Book Enclave, Jaipur, India, 2003. [3] D. N. Sujatha, K. Girish, K. R Venugopal, and L. M. Patnaik, An Integrated Quality-of-Service Model for Video on Demand Application. IAENG International Journal of Computer Science, 2007, 34, 1. [4] E-Learning Case Study, http://www.newman.ac.uk/students_websites/ ~m.m.friel/advcs.htm, Updated on 17 June 2004. [5] Fripp, John, Learning Through Simulations: A Guide to the Design and Use of Simulations in Business and Education, McGraw-Hill, 1993. [6] Huang, Cheng, Jin Li, and Keith W. Ross, Can Internet Video-on-Demand Be Profitable? SIGCOMM, 2007. [7] Jennings, Roger, Cloud Computing with the Windows Azure Platform. Indianapolis, IN: Wiley Publications, 2009. [8] Vecchiola, Christian, Suraj Pandey, and Rajkumar Buyya, High-Performance Cloud Computing: A View of Scientific Applications. Cloud Computing and Distributed Systems. [9] Velte, Anthony T., Toby J. Velte, and Robert C. Elsenpeter, Cloud Computing a Practical Approach, New York: McGraw-Hill, 2010. [10] P. Seethalakshmi and V. Ramachandran, RMI based Distributed Model for Media on Demand, Internet and Multimedia Systems and Applications, 2002 [11] Ping Zhang, Weizhong Liu, Xuecheng Zou, A Study of Video-on-Demand Learning System in E-learning Platform, International Conference on Computer Science and Software Engineering, 2008.