- Who: Microsoft
- What: Live streaming of the Winter Olympics at Sochi in February 2014 using Azure Media Services
- How: Microsoft used platforms contributing to TM Forum’s Open Digital and ZOOM programs, and used SDN and NFV to assign virtualized resources
- Results: Microsoft’s six broadcasting partners were able to support 61 million unique viewers with 6,000 hours of high-definition streaming, including setting a streaming event record of 2.1 million concurrent viewers for the semifinal of the U.S. versus Canada hockey game with no service affecting outages.
When the U.S. and Canada faced off in the men’s hockey semifinals during the Winter Olympics in February 2014, Microsoft and its broadcasting partners knew the streaming event would be big, but they didn’t know it would break records with 2.1 million concurrent viewers. Later a World Cup soccer match between Cameroon and Mexico in June set a new record for simultaneous viewers at 2.8 million. Microsoft Azure Media Services delivered video for both events flawlessly – in fact, there were no service-affecting outages reported during any of the 18 days of Olympic coverage.
Azure Media Services is a cloud service that enables an ecosystem of partners to collaborate and create media workflows leveraging first- and third-party services to deliver digital content to consumers on a variety of devices. The service is enabled by the Microsoft Azure Cloud Platform. Microsoft does not release revenue figures for individual services, but the company says Azure Media Services is one of the fastest-growing services on Windows Azure since it was launched in January 2013.
“Sochi was a highly visible, real-time event, and that means there’s no downtime tolerated,” says Jason Suess, Principle Program Manager, Microsoft. “We needed to be able to handle the scale of the Olympics with so much happening all at the same time. At peak, we were running around 15 simultaneous events, so that meant 15 HD [high-definition] streams running concurrently. From an uptime perspective, we needed redundancy all along the delivery chain.”
Sochi value fabric
The Sochi delivery chain was really more of a value fabric comprising many partners (see graphic below).
Here’s how it worked:
Six broadcasters worldwide (including NBC Sports in the U.S., NHK in Japan, CBC Canada, America Movil in Latin America and ARD/ZDF in Germany) received video streams from the International Broadcasting Center in Sochi. The live feeds were pushed through an aggregator, iStreamPlanet, which, in turn, pushed feeds into the Azure cloud.
As feeds came into the cloud, software load balancers (SLBs) on a triple-redundant available pool of virtual machines (VMs) ensured smooth ingestion processing. Microsoft Azure Media Services then enabled the creation and execution of defined work streams, leveraging a variety of first- and third-party services to enrich the digital content.
At egress, Microsoft Azure provisioned multiple VMs in the origin deployments, both to ensure redundancy and enough origin scalability to handle the highly variable loads. Here, virtualized SLBs ensured that the load was distributed across all the role instances and only to the instances that were healthy. From the Azure cloud, the video streams were sent to Akamai CDN, a content delivery network, and from there they were distributed through broadband service providers to end users’ mobile and desktop devices via an Adobe client.
Under the covers
“From a networking perspective, there’s a lot going on under the ‘platform-as-a-service’ covers that enables us to be very agile, efficient and elastic in our responses to rapidly changing conditions,” says Eric Troup, Chief Technical Officer, Worldwide Communications and Media Industries, Microsoft. “This case study highlights how Azure uses software-defined networking for agility and virtualized network functions for efficient assignment of virtualized compute, storage and networking resources. Azure is essentially an important way-station along a value fabric that ultimately extends out to anywhere a service provider streams traffic to consumers.”
It helps to think of Azure as a digital services enablement platform that sits in the middle of two ecosystems of partners, Troup says. On one side, Azure Media Services connects the partner companies that deal with the content itself – for example, the company that records the video, sends the live video steam, creates the video-on-demand content and preps for associated advertising and its partners like Deltratre, which inserts value-added player, game and sports statistics. On the other side, Azure Media Services enables streaming of the enhanced content to an ecosystem of service providers that deliver it to end users through cable, telecom or mobile broadband services.
“Azure also provides common management service APIs that have to do with provisioning and service assurance,” Troup explains. “These simple management interfaces are delivered by the Azure Management Portal, with partner onboarding, catalog, charging and settlement functions handled by the Microsoft Commerce Platform.”
Scaling peaks in Olympic time
To deliver the content from Sochi, Microsoft used virtualized compute nodes in the Azure cloud to facilitate live transcoding in parallel across multiple nodes. If one virtual machine went down, then the content was automatically distributed to a new one. As the load peaked – like for the hockey game – the whole environment had to be able to expand and contract automatically.
This is fairly easy to orchestrate inside one data center where there is tight control, but not when multiple clouds are involved. “It becomes more difficult when it’s multi-cloud and when you have different network controllers in control of different resources,” Troup explains. “For Sochi we had to do a lot of upfront planning to overcome the limitations. We’re hoping in the future to be able to do it more quickly through industry standard interfaces and APIs.”
That’s why Microsoft is participating in TM Forum’s Zero-touch Orchestration, Operations and Management (ZOOM) program and in a Catalyst project focused on multi-cloud SDN-NFV service orchestration. “We know how to do service chaining when it’s contained within the boundaries of a single network controller, but there isn’t a standard way to extend the control beyond two or more controllers. It requires some kind of higher level control and the industry hasn’t settled on what it should be.”
Microsoft at work in the Forum
Microsoft has contributed to and continues to reference much of TM Forum’s work. For example, the Microsoft Commerce Platform (MCP), which delivers back-office capabilities like customer care, fulfillment, provisioning, charging and billing for Azure Media Services (AMS), was built using the Frameworx suite of standards-based tools and best practices.
MCP is currently being updated into a consolidated billing platform for all Microsoft services such as Azure, Xbox Live and Office 365. In addition, the Microsoft Online Services MCP team is participating in further development of the TM Forum Digital Services Reference Architecture (DSRA – you can also find out more about the DSRA by downloading, free, the Quick Insights report, New services: Bridging the gap into the digital world).
The Azure Management Portal, part of Azure Cloud Platform, provides service assurance capabilities for Azure users, and its capabilities are a derivative of work done by Microsoft Communications Sector on a previous offering called the Connected Services Framework, which delivered the concept of a ‘Well Enabled Service’ (WES). WES is a service that exposes one or more management interfaces, and it was contributed to the Forum’s work on Service Delivery Framework (SDF), which has evolved into the DSRA.
Microsoft contributions also helped define TM Forum’s Simple Management API. Today the Azure Management Portal exposes business and performance metrics that are closely aligned to the Simple Management application program interface. The interfaces enable a multi-cloud end-to-end view of how services are performing and were used to support the management of the streaming of the Summer Olympics in 2012 from London, and the Winter Olympics and soccer World Cup in 2014.
Another significant Microsoft contribution to TM Forum includes the company’s Multi-Cloud Service Delivery and End to End Reference Architecture paper, which helped jump-start the Forum’s Multi-Cloud Management Accelerator, also part of the DSRA. According to Microsoft, AMS is an example of the multi-cloud vision at work, blending IT cloud services such as media encoding and network function cloud services such as content delivery networks and SDN and NFV into a seamless open digital delivery platform.
Back in 2008, TM Forum hosted the Digital Content Encounter with a series of enhanced Catalyst projects running over four six-month cycles to identify and standardize the crucial ‘management touch points’ between operational systems and business processes of all the players in the extended value chain for delivering converged services end-to-end. AMS is a commercially available cloud service that delivers on many of the goals that were the focus of the Content Encounter.
A depiction of the reference architecture is shown below:
[Image: Olha Insight /Shutterstock.com]