Paper Abstract and Keywords |
Presentation |
2020-03-06 10:10
8K ultra-high-definition video stream server using virtual machines on KVM Tetsuya Ikeda, Katsuhiro Sebayashi, Kouta Aoki, Mitsuru Maruyama (KAIT), Hiroyuki Kimiyama (DIT), Takashi Kurimoto, Shigeo Urushidani (NII), Takashi Kohno (NTT-TX) NS2019-224 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
With the formulation of the SMPTE ST-2110 standard, the editing of broadcasting contents and the adoption of IP in the storage and distribution environment are progressing along with the expansion of the cloud service market and the virtualization of network functions. This study is aimed at creating a distributed storage and distribution system for an 8 K uncompressed video that can flexibly use the free resources of the cloud. In this study, the issues and tuning methods for stable delivery of 24 Gbps 8 K uncompressed video are presented by linking eight VMs with the required transmission performance of 3 Gbps. In the basic experiment, the performance of the reading video from the shared storage by the video distribution application was not stable, and the issue remained that multiple VMs did not operate stably on one physical server. Therefore, SR-IOV was enabled and a file sharing protocol was selected to improve the protocol processing performance of the VM. In addition, the OS was changed on the VM and the number of vCPU cores and the amount of memory for stable operation of multiple VMs were reduced. As a result, pinning the vCPU 4core to each VM aided in allocating 16 GB of memory. Additionally, virtually splitting the 10 Gbps-NIC, and arranging the two VMs on the physical server helped stably link the 8 K uncompressed video. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
8K / 8K-UHD video transmission / KVM / NFV / SINET5 / virtualization technology / / |
Reference Info. |
IEICE Tech. Rep., vol. 119, no. 460, NS2019-224, pp. 259-264, March 2020. |
Paper # |
NS2019-224 |
Date of Issue |
2020-02-27 (NS) |
ISSN |
Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
NS2019-224 |
|