9
MONETIZING THE EDGE WITH CONTENT DELIVERY NETWORKS

Monetizing the dgee with Content delivery networks

  • Upload
    others

  • View
    5

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Monetizing the dgee with Content delivery networks

Monetizing the edge with Content delivery networks

Page 2: Monetizing the dgee with Content delivery networks

Edge computing is increasingly prevalent and will become even more so as the use of 5G and Internet of Things (IoT) devices enter the mainstream. By locating key data processing functions closer to the edge, there’s no need to transmit data to the cloud and back for analysis, so latency is less of an issue. Also, since edge data centers are increasingly self-sufficient, it is still possible to keep services up and running even when other parts of the infrastructure are down.

Sixty nine percent of service providers already have plans to deploy multi-access edge computing (MEC)i. Yet the majority are still trying to figure out the golden use case that will monetize their investment. There’s a lot of talk about cloud gaming, 360-degree virtual reality (VR), and immersive media, but the real use case driving return on investment (ROI) today is content delivery networks (CDNs).

In this eguide you’ll discover why CDNs are becoming increasingly lucrative, the key challenges associated with CDN delivery, and how an integrated, end-to-end computing infrastructure optimized for media workloads can address these.

introdUCtionContentsIntroduction ..................................................................... 2

Why are Content Delivery Networks important? ..... 3

What are the key CDN challenges? ............................. 4

What are Intel's solutions for CDN? ............................ 5

Memory, bandwidth and storage .................................... 5

Accelerators ............................................................................ 6

Ecosystem acceleration ..................................................... 6

Open Visual Cloud ............................................................... 6

Intel® Select Solutions ....................................................... 7

Conclusion ........................................................................ 8

Page 3: Monetizing the dgee with Content delivery networks

why are Content delivery networks iMportant?Media is transforming. Thanks to our insatiable appetite for live and on-demand video, the global video streaming market will be worth USD 102B by 2023ii. Consumers expect this web and video content anytime, anywhere and on any device. And as the content gets richer, more localized, personalized, interactive and immersive, the pressure is on for service providers to deliver more than just ‘media’ content to their customers; they must deliver Visual Cloud experiences. These range from media processing and delivery, through to new services such as media analytics and dynamic ad insertion. CDNs enable service providers to deploy these services at the edge of the network, which means they can deliver richer content to widely distributed, and often mobile, consumers.

Page 4: Monetizing the dgee with Content delivery networks

1. Web content: With rapid rise in encrypted web traffic, more businesses are using encryption as the primary method of securing web content. CDN providers are required to match the growing levels of data from secure sockets layer (SSL)/ transport layer service (TLS) encrypted websites. This requires cryptography to be done by the server’s CPU located in CDN infrastructure. Increasing the throughput requires CDN providers to upgrade those servers to the latest processor technology.

2. Long-tail content: With growth in over-the-top (OTT) services and

video traffic, large Internet traffic spikes are everyday occurrences. To overcome this challenge, many CDN providers are transitioning to higher connectivity of 100G per cache node to scale their node level performance. For long-tail video content such as video-on-demand (VOD) and cloud digital video recorder (DVR) usages, this requires CDN providers to transition to high throughput storage such as non-volatile memory express (NVMe)-based storage devices to be able to fully utilize the available network throughput.

3. Live streaming: One of the other challenges faced by CDN providers

is with live linear content streaming. The live streaming video market is growing at an alarming rate and since the content is immediately distributed to the end users, large memory capacity is needed to buffer each of the many independent streams a CDN server needs to handle. Unfortunately, large capacity dual in-line memory modules (DIMMs) are expensive.

These key use case specific challenges are shown in figure 1.

what are the key Cdn Challenges?

Cdn Use Case Challenges

Transparent caching

Long-tail content

Live streaming and hot VOD caching

• Need more cores for security encryption/crypto acceleration

• Maximizing available throughput

• Overprovisioning of capacity to meet throughput demand

• Failure rates with HDD

• Exponential rise in live streaming traffic

• Lower glass-to-glass latency

• Need for more memory, low latency caching

SSD

SSD

Memory

HDD

Figure 1: Key CDN challenges.

Page 5: Monetizing the dgee with Content delivery networks

Intel offers comprehensive, interoperable end-to-end media solutions to help service providers better meet the aforementioned CDN challenges so that they can better serve their existing customers, as well as extend their CDN to support emerging edge cloud use cases.

MeMory, bandwidth and storage

• Intel® QLC 3D NAND SSDs are ideally suited for high-capacity, high-volume use cases such as VOD and cloud DVR

• Intel® Optane™ SSDs provide extremely high reliability and performance compared to hard disk drives (HDDs) and standard SSDs. These drives are best suited for hot content use cases

• Intel® Optane™ DC persistent memory provides a bigger memory footprint at a low total cost of ownership (TCO) and faster access to storage by bringing data closer to the CPU. This technology helps service providers meet growing demand for latency-sensitive use cases such as live and linear streaming

• 2nd generation Intel® Xeon® Scalable processors support demanding CDN workloads, with two 48-core sockets, 12 memory channels per socket, high-speed interconnect and cache optimizations, and balanced I/O to meet the demands of high-performance storage solutions

• Extending intelligence to the edge, Intel® Xeon® D processors support high-density, single-socket network, storage and cloud edge computing solutions with a range of integrated security, network and acceleration capabilities

what are intel’s solUtions for Cdn?aCCelerators• High-density accelerators like those from Celestica – see

figure 2 – can help to maximize performance on edge nodes

• First, the Intel® Visual Cloud Accelerator Card – Analytics (VCAC-A) is optimized for the highest density offload acceleration for use in network edge servers. The new acceleration solution combines both on-card decode and video inference acceleration, along with high efficiency, FP16 precision that can manage 24 streams at 1080p 30FPS, up to 144 streams per server with six cards

• Second, the Intel® Visual Cloud Accelerator Card – Render (VCAC-R) is optimized for cloud gaming, graphics, VR, and virtual desktops. With high-performance rendering and

encoding, low latency, competitive 3D performance, and multiple Windows games per card, VCAC-R can handle four Windows game streams at 1080p 30FPS, and at 720p 30FPS can handle 20 Android game streams

• As an alternative to upgrading to the latest processor technology to manage encryption throughput for secure web content, Intel® QuickAssist Technology (Intel® QAT) cards help to offload cryptography from the CPU to expand its SSL/TLS throughput, while being cost efficient. For inline SSL/TLS processing, CDN providers can further utilize the Intel® FPGA Programmable Acceleration Card (Intel® FPGA PAC) N300

Figure 2: Visual Cloud Accelerator cards from Celestica.

rendering analytiCsDensity-OptimizedPerformance/TCO

Workload OptimizedLow Power

• High 3D performance at competitive TCO

• Low latency rendering and encoding

• Multiple windows games per card

• Optimized for high-density edge analytics

• Up to 24 real-time streams on board

• Intel® Movidius™ Myriad™ X VPU

Page 6: Monetizing the dgee with Content delivery networks

eCosysteM aCCeleration

• Intel is working to accelerate the CDN ecosystem through collaboration with communication and cloud service providers; workload characterization and performance turning; and hardware and software optimizations with original equipment manufacturers (OEMs), systems integrators (SIs) and independent software vendors (ISVs) – see figure 3

open visUal CloUd • To help strengthen the ecosystem and

provide ready access to the building blocks and pipelines for cost-effective Visual Cloud innovations, Intel has set up the Open Visual Cloud project. It provides availability of high performance, high quality, open source, validated building blocks across encode, decode, inference, and rendering – see figure 4, as well as reference pipelines that support visual cloud workloads. The goal is to minimize barriers to innovation for quickly and easily creating and monetizing Visual Cloud services. Support for familiar industry standard frameworks use the larger open source community and include media (FFMPEG and GStreamer), AI (TensorFlow, Caffe, MXNet, ONNX, Kaldi), and graphics (OpenGL, DirectX) Figure 3: Characterization and optimization through a collaborative process.

CUstoMer Collaboration

hw and sw optiMizations

workload CharaCterization and perforManCe tUning

Replication of customer CDN environment in Intel Labs (3 labs located in AZ, MA, PRC)

Join testing and optimization with OEMs, SIs and ISVs

Engagements with telco CDN providers, OTT providers, next wave CSPs

Validated solution with CDN stack improvements (Linux, NGINX, ATS)

Guidance on best practices for CDN deployments (targeted white papers, sizing guides, case studies

and detailed documentation)

Performance and security optimizations

Page 7: Monetizing the dgee with Content delivery networks

intel® seleCt solUtions• Intel® Select Solutions for Visual Cloud Delivery

Network are workload optimized solutions that have been verified by Intel to achieve benchmarked results and meet the rapidly emerging needs for CDNs. Delivering scale, flexibility and high performance, these solutions are built by Intel® Network Builders partners on 2nd gen Intel Xeon Scalable processors and are designed to accelerate visual services at the edge of the network.

To find out more, read the solution brief: Intel Select Solutions for Visual Cloud Delivery Network Figure 4: Intel’s validated building blocks for the Visual Cloud.

inferenCe: intel® distribUtion of openvino™ toolkit

enCode/ deCode: sCalable video teChnology (svt) for hevC and av1

rendering: intel® one api rendering toolkit

• Enables deep learning inference at the edge

• Speeds time to market for computer vision and media analytics applications via a library of functions and pre-optimized kernels

• Provides media and visual cloud developers with high-performance encoder libraries

• Helps to achieve the best-possible trade-offs between performance, latency and visual quality

• Offers a flexible platform as an alternative to dedicated graphics accelerators

• Supports efficient delivery of high-performance, high-fidelity video graphics, visualization and rendering

Page 8: Monetizing the dgee with Content delivery networks

Service providers are turning to edge computing to address concerns around latency and downtime as use of 5G and IoT become more widespread. But many service providers have struggled to find a use case to monetize the edge until now. As our insatiable appetite for video anytime, anyplace and on any device grows, the provision of CDNs at the edge presents service providers with huge revenue-generating potential. However, CDN delivery does not come without its challenges with regards to infrastructure performance, throughput and memory. Service providers can meet these challenges head on with integrated, interoperable end-to-end media solutions from Intel.

Find out more:

Visit www.intel.com/visualcloud

Intel Optane DC Persistent Memory for CDNs: https://www.intel.com/content/www/us/en/products/docs/memory-storage/optane-persistent-memory/content-delivery-network-video.html

ConClUsionChina teleCoM laUnChes virtUal CloUd delivery network (vCdn) testing on edge data Center

• The China Telecom Guangzhou Research Institute conducted an edge data center vCDN proof of concept (PoC) using an Intel® Select Solution for Network Functions Virtualization infrastructure (NFVi) and Intel® Visual Computer Accelerator (Intel® VCA)

• Its objectives were to evaluate the support capacity of edge data centers for new service applications, verify whether the Intel Select Solution for NFVi-based server satisfies the requirements of edge data centers and next-generation central offices (NGCOs), and to evaluate the video encoding and transcoding acceleration performance of Intel VCA on a vCDN

• The PoC proved that edge data centers can effectively drive the implementation of NFV-based core network functionalities to provide users with a better service and experience

• The Intel Select Solution for NFVi in an edge data center provides the necessary computing, storage, and network capabilities, and meets security, reliability, and high availability requirements. While Intel VCA provides outstanding performance in video acceleration and accommodates vCDN in edge data centers

To find out more, read the case study: China Telecom Verifies Edge Implementation of NFV-based Core Network Functions

Page 9: Monetizing the dgee with Content delivery networks

i Heavy Reading, September 2019 “Transforming the Edge: The Rise of MEC” https://www.intel.com/content/dam/www/public/us/en/documents/white-papers/the-rise-of-multi-access-edge-computing-paper.pdf

ii Kenneth Research, October 2019 “Global Video Streaming Market Analysis & Forecast, from 2017 to 2023” https://www.marketwatch.com/press-release/video-streaming-market-industry-insights-by-growth-emerging-trends-and-forecast-by-2023-2019-10-01

Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more complete information visit www.intel.com/benchmarks

Configurations: CPU: Intel® Xeon® Gold 6252 “Cascade Lake” Processor MEMORY: 192 GB DDR4 DRAM Memory + 1536 GB Intel® Optane™ DC persistent memory compared with CPU: Intel® Xeon® Gold 6252 “Cascade Lake” Processor MEMORY: 1536 GB DDR4 DRAM Memory. Testing carried out by Intel on 13th March 2019.

Performance results are based on testing as of the date set forth in the configurations and may not reflect all publicly available security updates. See configuration disclosure for details. No product or component can be absolutely secure.

Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration. No product or component can be absolutely secure. Check with your system manufacturer or retailer or learn more at intel.com

Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy.

Your costs and results may vary.

Notice Revision #20110804

Intel, the Intel logo, and other Intel Marks are trademarks of Intel Corporation or its subsidiaries in the U.S. and/or other countries.

Other names and brands may be claimed as the property of others.

© Intel Corporation 0220/FP/CAT/PDF 342450-001EN