Media is evolving rapidly; it’s no longer just about watching Netflix on the living room TV or on our PCs. Today, at Intel’s Data Centric Innovation Summit, we discussed how we are surrounded by rich, visual user experiences that include broadcasting live events, cloud gaming, public safety, AR/VR training, highly complex graphically rendered movies, and more.
A simple label used at Intel to describe this set of experiences is ‘Visual Cloud’, representing a set of diverse visual workloads (e.g., media processing and delivery, media analytics, immersive media, cloud gaming, and cloud graphics) delivered by a variety of service providers. Moving beyond simple streaming into broader Visual Cloud means that the demands for solutions include not just decode and encode functions, but increased integration of Artificial Intelligence and Machine Learning through inferencing as well as rendering capabilities.
Professional visual content has been processed in the cloud and consumed remotely for many years, but the ubiquity of cameras is also driving a steep ramp of user-generated content being uploaded for processing in the cloud. Cisco’s Visual Networking Index 2017 indicates that, by 2021, approximately 80% of the wireless and wireline traffic will be video, and in a recent article, Huawei stated that "some 300 hours of video is uploaded to YouTube every minute, half of which is viewed on mobile devices…75% of Facebook video browsing is performed on smartphones.1"
Business Needs, User Experience Shape the Ideal Infrastructure
Clearly consumers demand rich content and experiences on their terms, any time, on any device and, without delay—regardless of the limitations of their own device or network. What does this mean for the service providers?
Multi-Generational Codec Flexibility: First, the service provider platform must have the flexibility to support multiple generations of codecs, screen formats, network topologies, and localized encoding optimizations on thousands of devices. This makes scalable, standards based architectures an essential requirement as the foundation for their solutions.
Cloud-Like, Virtualized Workload Capability: Second, many service providers find the greatest operational efficiencies from supporting a diversity of workloads on the same platform, as well as multiple simultaneous user sessions. This means a platform must be capable of multi-tenancy and dynamically shifting workloads to efficiently adjust to customer demand profiles.
Platform advancements in virtualization and container support, as well as network advancements such as NFV and SDN, can be instrumental enabling technologies. In addition, general purpose, scalable processors that are capable of performing diverse and complex functions can be highly cost effective over multiple service generations relative to dedicated fixed function hardware.
Latency That Meets Service Level Agreement: Third, service providers must deliver user experiences that are in line with demands of target applications and usage models. Network latency plays a crucial role in end-user experience, requiring the service providers to carefully weigh their workload placement decisions.
While traditional media streaming is increasingly sophisticated in managing latency through local analytics and use of CDNs, applications such as interactive cloud gaming and real-time AR/VR experiences are extremely sensitive due to the rapid two-way information exchange across the network. Placing the Visual Cloud processing capabilities closer to the edge and leveraging the strength of advanced networks, such as 5G, will increasingly ensure a user experience in line with customer expectations. Based on their unique business model and service type, each service provider must determine how to best balance intelligence in the cloud relative to the availability of cloud-ready networks and processing closer to the edge.
Bandwidth That Delivers the User Experience: Fourth, similar to network latency, network bandwidth, the amount of data delivered at once, can also make or break a service provider’s business. A company like Netflix may not be as latency sensitive but certainly needs to secure sufficient network bandwidth that allows them to continue adding subscriptions at costs that ensures growth in net profits. For someone like the NFL, broadcasting a live sporting event can quickly ramp from several thousands to several millions streaming in real time, and they must be prepared to respond.
Cost-Effective Infrastructure: Finally, service providers must deliver rich services and user experiences at a total cost of ownership that aligns to their core business model, without sacrificing fastest time to market.
The recipe is complex with service providers who are working to predict and keep pace with consumer demands and optimize investments against difficult solution tradeoffs with long-lasting effects. In order to survive competitive pressures and ensure longevity of their investments, service providers will need a platform that is flexible, scalable, and based on a cloud architecture with modernized networks. This requires a fundamental shift of underlying infrastructure innovation towards a focus on workload-optimized systems architecture.
Intel Takes a Wide-Angle View
Intel is enabling innovation in Visual Cloud through a comprehensive end-to-end approach. We offer a rich portfolio of processors with scalable performance, ranging from powerful general purpose Intel® Xeon® processors to hardware accelerators such as Intel® Movidius™ VPUs, Intel® FPGA, and the Intel® Visual Compute Accelerator card.
Intel® software libraries and tools scale the entire range of Intel® processors, ultimately supporting the four pillars of Visual Cloud workloads: Render,inference,encode, and decode. This allows service providers to innovate with the flexibility of a general purpose Intel® Xeon® processors to deliver multiple workloads on a single platform architecture, and use software optimizations to quickly and cost effectively accelerate target workloads.
Intel recognizes the crucial role of network in delivery of visual workloads and is continuing to aggressively drive network advancements with key technologies, such as SDN/NFV and 5G. Intel’s investment in a silicon foundation across compute, memory, storage, and network, and a system’s approach to workload optimized infrastructure has created unmatched innovation for the drivers of data-centric infrastructure capability.
This is an exciting time to be a part of shifting what was once seen as science fiction into the delivery of compelling visual experiences. You can already see the shift happening in the industry. Just turn on your TV…..or your phone…..or your gaming PC….or your VR headset. You get the picture (pun intended).
I hope that you join us on this journey of innovation and opportunity in the Visual Cloud! I invite you to stay tuned for further announcements around Intel innovation at upcoming events like SIGGRAPH and IBC.
Check out all the news from Intel’s Data Centric Innovation Summit here.