1st International Workshop on Cloud Education Environments (WCLOUD 2012) Render on the cloud: Using Cinelerra on virtualized infrastructures I. Zablah A. Garcia-Loureiro, F. Gomez-Folgar and T.F. Pena Sistema de Difusión de Radio y Televisión Centro de Investigación en Tecnoloxı́as da Información (CITIUS) Universidad Nacional Autónoma de Honduras University of Santiago de Compostela Tegucigalpa, Honduras Santiago de Compostela, Spain mrzablah@unah.tv (antonio.garcia.loureiro, fernando.gomez.folgar, tf.pena)@usc.es Abstract—Nowadays, the learning-teaching processes are The cloud technology can be defined as some kind of being supported by the use of new technologies, including parallel and distributed system [4], conformed by a lot email, chat, online conferencing, online activities and video- of interconnected Virtual Machines (VM), or guest sys- conferencing. With the revolution of high definition television, video producers require more efficiency in the production and tems, providing dynamically computational resources on post-production tasks. There are several software available demand as a unified one. They are based on a Service for this purpose including commercial solutions, often very Level Agreement (SLA) [5]. This technology is growing expensive, and open-source ones, such as Cinelerra. very fast as well as the computer resources that support This paper proposes a cloud infrastructure for using Ci- it, specially the Service Oriented Architecture (SOA) [6] nelerra, a community developed version of non-linear video editor, and how educational institutions can use this. The main and the virtualization technology [3], using both hardware idea is to reuse its computational power for editing or creating and software resources. The virtualization technology is the educational videos without the need of acquiring a dedicated cornerstone of the cloud, as well known as Infrastructure as hardware infrastructure, employing non dedicated resources, a Service (IaaS) [7]. The cloud technology can be used with such as the computer labs, or desktop computers to fix the different objectives. In our case we are interested to know most common time consuming problem: the rendering. The performance of the proposed infrastructure is also presented the advantages of use it on video rendering process using in this paper. the Cinelerra [8] application on a virtualized environment provided by a hypervisor layer. Keywords-Cinelerra; rendering; cloud; broadcast; This paper is organized as follows. The section II de- scribes the implementation of the proposed infrastructure I. I NTRODUCTION used to install Cinelerra and how use it to improve the ren- Recently, technologies [1] such as email, chat, online dering process. The section III describes two cases of study conferencing, online activities and videoconferencing were for the proposed infrastructure. The Section IV includes incorporated to support teaching and learning processes. the performance evaluation of the proposed infrastructure Today we live the biggest revolution on computing, mul- performing several rendering tests. Finally, the conclusions timedia and TV, since the invention of color broadcast of this paper are drawn in section V. in the beginning of fifties decade [2]. Currently, teachers and students can watch videos on a variety of mediums, II. P ROPOSED I MPLEMENTATION from mobile phones, computers and high definitions screens. The proposed infrastructure for executing Cinelerra, These changes goes on the hand with the evolution of the shown in Fig. 1, is composed by several types of compo- information technology. They used the computing power of nents: physical compute nodes, a virtual master node, virtual the new computers to create videos of superior quality and render nodes and a Network File System (NFS), shared by complexity from a bunch of source feeds. the virtual cluster. It is well-known rendering is a time consuming task. The physical compute nodes are an Intel Core I7-2820QM Usually, the companies and training centers use expensive processor with a clock speed of 2.30 GHz and 8 GB of resources to reduce the edition time necessary to prepare DDR3 RAM, with the VirtualBox [9] 4.1.14 hypervisor the videos. In an educational environment, it could be better installed. This processor reports eight cores (with hyper- to create high quality videos without the need of incurring threading enabled). The master and the render nodes are Vir- in license cost and without acquiring expensive dedicated tualBox VMs with two cores per VM, with 1 GB of RAM, hardware resources. This paper focus on how it is possible taking advantage of the hardware virtualization extensions. to reuse the existing computer hardware of educational The virtual cluster infrastructure (virtual master and virtual institutions, such as schools, colleges and faculties to create compute nodes) uses Ubuntu 12.04 LTS 64 bits as guest educational videos, by implementing cloud and virtualiza- operating system, employing a Gigabit Ethernet interface. tion technology [3]. The NFS stores input and output video resources. 28 1st International Workshop on Cloud Education Environments (WCLOUD 2012) Figure 2. Use of Cinelerra at UNAH TV station. Figure 1. Schematic view of the implementation. In the Editing & Production department, the editors, III. C ASES OF S TUDY equipped with powerful workstations, access to the hierar- chical Video Storage to take the required resources that will Two cases of study were proposed as examples of use of be incorporated into the Cinelerra video project, including the infrastructure described previously. The first one is the sounds, videos, images, and production scenarios. Once the implementation of Cinelerra over the infrastructure avail- video is edited, is ready for starting the rendering stage able at the National Autonomous University of Honduras employing the computational power available at the Cloud (UNAH). The second one is using Cinelerra in a training infrastructure that is composed by non-dedicated hardware center. resources. The rendering stage ends with the creation of a A. Cinelerra in the public UNAH TV channel final version of the ready to air TV program. The results of The infrastructure proposed previously can be imple- the rendering process are stored in the Video Storage, which mented at UNAH in order to help the process of producing can be taken by the operators of the Play Out department television content for broadcast on the public TV channel to be transmitted and make them available to final users. owned by the university, reducing the required time to obtain Those contents can be also accessed by the academic staff the final content for airing. This infrastructure takes advan- from the classroom using the existing university network, or tage of the virtualization technology and the Infrastructure to the general public employing Internet streaming. as a Service (IaaS) paradigm, allowing reusing the hardware B. Cinelerra as a tool in educational environments available at UNAH for multiple applications. Fig. 2 shows the work-flow that must be follow with the This scenario proposed the use of Cinelerra as a tool in main objective to incorporate the advantages of rendering educational environments employing a cloud composed by on the cloud. As shown in the figure, at UNAH TV station non dedicated hardware resources. In this case, the cloud all the video and TV production starts with the introduction is employed as a rendering queue where the projects will of the media content into the broadcasting system through be processed like in a batch system to get the ready to air a common point called Input Resources or Ingest. Here videos. This case of study is shown in Fig. 3. As we can see, employing a serie of procedures, videos from a camera, students create non-linear edition projects using Cinelerra in studio, satellite, DVD, tape, etc. are converted in computer their workstations. Source videos, transitions effects, and the video files accompanied with a standardized metadata files additional necessary media compose those projects. When that describes their content and properties. the edition process is finished, the project is stored in As a result of the previous process, in first place, the video the Master Node. This element shares the media directory files are stored in the Video Storage, that is composed by employing the Network File System (NFS) protocol and has a group of NFS servers configured with redundant arrays also Cinelerra installed as a render queue that manages the of disk drives. In second place, the metadata is stored in jobs pending of being rendered. These jobs are dispatched a database server. The Video Storage can be organized in to the Virtual Render Nodes in which the ready to air video hierarchical levels to distinguish videos of different origin, is created. The Virtual Render Nodes are Cinelerra enabled source, resources, media support and the edited ready to air VMs that mount the NFS directories shared by the Master videos. Node. The virtual Render Nodes can be executed in the 29 1st International Workshop on Cloud Education Environments (WCLOUD 2012) Figure 4. Performance with 40 seconds of video at 1080/60p. employing two cores each one. The purpose of the tests is to know how much time is needed in every configuration and Figure 3. Use of Cinelerra by students with virtual render nodes located how the render process could be beneficed or penalized. on remote computers on other rooms. When several nodes are used, we set the option in Cinelerra that allows to automatically dividing the job in several parts. Cinelerra itself splits the jobs trying to give computer laboratories of the educational institutions. The the same number of render parts to all machines (including video created by the rendering process is finally stored in the master). In this way, the number of files generated scale the NFS shared directory of the Master Node. Notice that with the number of nodes employed (from one when only the student’s workstations, the Master Node and the Virtual the master rendering, two, four, six, eight up to sixteen). This Render Nodes must be accessible and interconnected by a method avoids that some rendering nodes were unused. communications network. For the first test, the video of 40 s at 1080/60p, we can The main goal of this scenario is the possibility to reuse see the obtained results in Fig. 4. The master needed more the idle computational power available in the computer labs time to finish than the other configurations, except when elastically, as they are available, to reduce the necessary time we used the master with one node and sixteen output files to get the rendering process finished. that required 287 s, the worst performance in this test. The best time was obtained employing the master and two nodes IV. P ERFORMANCE E VALUATION using eight files, requiring 125 s only. Employing the same In this section we included the performance analysis combination with four files the time required was 134 s. of Cinelerra over the proposed infrastructure, as described Using the master and one node, the best result was quite the previously. same using two and four output files; this was 153 s, this is Cinelerra was used for editing and rendering video on res- 28 s more than the best result of all the present test. olutions of 720p and 1080i, with input/output streams used Our second test was a project with an output video of for high definition (HD) television. As a common standard 1800 s (30 minutes) at a standard resolution of 720p. In this for audio/video the codec MPEG4 [10] was employed. It has project we joined some feeds and added transitions between been on video editing world from last years of 90s decade. every resource, applying basic effects. The rendering process To evaluate if the proposed cloud infrastructure could took 6487 s for the master. This was the worst result on the be considered as a good option for rendering, we prepare present project. After adding one node to help the master, the some tests. The first one was an output video of 40 seconds efficiency was improved. The best result in this combination length with 1080/60p resolution. The second one was a video was 3605 s, allowing Cinelerra dividing the output render of 30 minutes at 720p. The third one was a video of 30 job in six files. After adding a second node, the shortest and minutes at 720p, composed by thirty feeds of one minute best time was obtained employing eight output files that took without transitions or effects. We run the render tests only 3072 s only. The results are depicted in Fig. 5. Because the in the master with two cores, later with one and two nodes, previous video project are not homogeneous and contains 30 1st International Workshop on Cloud Education Environments (WCLOUD 2012) Figure 5. Performance with 30 minutes of video at 720p. Figure 6. Performance with 30 minutes of video at 1080/60p. Shortest bars, better performance. several types of effects and transitions, which mainly differ in complexity, we consider to be very interesting to know how the infrastructure proposed would behave to complete we want to render very small videos, the computing capacity a project in which resources were homogeneous, allowing available on workstation used by the students or editors will the nodes performing similar computational tasks. The third be sufficient to successfully complete the jobs. test, as shown in Fig. 6, was prepared employing 30 feeds It is important to mention that the proposed solution, one minute of video at 1080i without employing transitions. based on cloud rendering, will be helpful in the process In this case, the better time was obtained with the master of creating multimedia for the TV station of the National and two nodes, employing sixteen output files that required Autonomous University of Honduras, Therefore, we can 1226 s The result is very similar employing eight files. The provide to end users, in a short period, large amount of render using the master required 2614 s, the worst elapsed quality television videos and multimedia using the idle time in the present measures. Working with the master and computing power available. one node, the best time was 1477 s, using with sixteen files. The most important aspect of improving the rendering We got similar results when this configuration is used with process is to have a greater chance to train better technical eight output files as the difference was 24 s only. experts and to create fastest multimedia content, therefore, V. C ONCLUSION the students will have more computational resources to The use of cloud technologies to create high quality develop their projects and ideas, but without incurring in videos in an educational environment is feasible without the cost of acquiring a dedicated infrastructure. the need of incurring in license cost and without the need of acquire expensive dedicated hardware resources. This ACKNOWLEDGMENT paper analysed a cloud infrastructure for using Cinelerra, Part of this job was funded through the Fiduciary Fund a community developed version of non-linear video editor. of the Japan’s government - UNESCO with the program As shown, this software could be used and employed by Keizo Obuchi 2011-2012, and by FEDER funds and Xunta educational institutions for teaching video techniques and to de Galicia under project 09TIC001CT, contract 2010/28, and create educational material. The main goal is to use, in a by Spanish Government (MCYT) under project TEC2010- more efficiently way, the computational resources managing 17320. them as a virtual cloud infrastructure for rendering purposes. This avoids the need of acquiring a rendering dedicated hardware infrastructure allowing to reuse existing computer R EFERENCES labs or desktop computers making them available to the [1] Schar, S.G. and Krueger, H.; Using new learning technologies students where they can easily render their videos. with multimedia. IEEE Multimedia, Number 3, Volume 7, The infrastructure and use cases proposed provide better Pages 40-51, 2000. DOI:10.1109/93.879767, http://ieeexplore. performance when working with large projects. Therefore, if ieee.org/lpdocs/epic03/wrapper.htm?arnumber=879767l 31 1st International Workshop on Cloud Education Environments (WCLOUD 2012) [2] WIRED: The First Color TV; March 25, 1954: RCA TVs Get the Color for Money. http://www.wired.com/science/ discoveries/news/2008/03/dayintech 0325 [3] Bhardwaj, Sushil, Leena Jain, Sandeep Jainl; CLOUD COM- PUTING: A STUDY OF INFRASTRUCTURE AS A SERVICE (IAAS). International Journal of Engineering and Information Technology, Pages 60-64, 2010. [4] J. J. Rehr, J. P. Gardner, M. Prange, L. Svec, and F. Vila. Scientific Computing in the Cloud, December - 2008. http: //arxiv.org/abs/0901.0029 [5] Buyya, Rajkumar and Garg, Saurabh Kumar and Calheiros, Rodrigo N.; SLA-Oriented Resource Provisioning for Cloud Computing: Challenges, Architecture, and Solutions. 2012. arXiv:1201.4522, http://arxiv.org/abs/1201.4522 [6] Hassan, Qusay F.; Aspects of SOA: An Entry Point for Starters. February, 2012. arXiv:1202.6623, http://arxiv.org/ abs/1202.6623 [7] Ambrust, Michael et al.; Above the Clouds: A Berkeley View of Cloud Computing. EECS Department, University of California, 2009. http://www.eecs.berkeley.edu/Pubs/TechRpts/ 2009/EECS-2009-28.html [8] Cinelerra-CV: Video editing for Linux. http://cinelerra.org/ [9] VirtualBox. https://www.virtualbox.org/ [10] MPEG-4 - The Media Standard. 2002. http://www.m4if.org/ public/documents/vault/m4-out-20027.pdf 32